Lasso 8 on Apache

classic Classic list List threaded Threaded
15 messages Options
Reply | Threaded
Open this post in threaded view
|

Lasso 8 on Apache

Jon Harris-2
Hi

We run Lasso 8 on a Windows VM under Apache 2.4.

At peak times, (which is generally early evening) the site really slows down. We have been through our code adding "cache" tags everywhere:

Cache(-Name="headerCache", -Expires=120) ;
// the code....
/cache ;
(etc, etc)

- and the overall performance has improved.

However, it is still lumpy (we are seeing a number of 408 errors in the apache logs) We have tried adding additional processors and RAM to the VM - but this hasn't made any noticeable difference. As Lasso 8 is only 32-bit (although the connector is 64-bit) there doesn't seem any point allocating the server any more RAM, than the 8Gb already allocated to it.

Looking at httpd.conf, I noticed there are a number of cache and proxy modules, most of which are commented out on our (default) installation.

LoadModule cache_module modules/mod_cache.so LoadModule cache_disk_module modules/mod_cache_disk.so LoadModule cache_socache_module modules/mod_cache_socache.so LoadModule proxy_fcgi_module modules/mod_proxy_fcgi.so LoadModule proxy_hcheck_module modules/mod_proxy_hcheck.so LoadModule proxy_html_module modules/mod_proxy_html.so LoadModule proxy_http_module modules/mod_proxy_http.so LoadModule proxy_scgi_module modules/mod_proxy_scgi.so LoadModule proxy_uwsgi_module modules/mod_proxy_uwsgi.so LoadModule proxy_wstunnel_module modules/mod_proxy_wstunnel.so

Do any of these modules make an difference to the performance of Lasso?

Also, we have a dozen or so rewrite rules in our .htaccess.  Is it worth trying to ditch .htaccess and write those rules into the httpd.conf? My understanding is that for every file request Apache has to run the url through the .htaccess and as result it can run faster with without one.

Beyond this, without major rewriting, would a front-end proxy server be the only viable solution.

I'd be grateful for any useful advice.

Regards
Jon Harris


#############################################################

This message is sent to you because you are subscribed to
  the mailing list Lasso [hidden email]
Official list archives available at http://www.lassotalk.com
To unsubscribe, E-mail to: <[hidden email]>
Send administrative queries to  <[hidden email]>
Reply | Threaded
Open this post in threaded view
|

Re: Lasso 8 on Apache

Bil Corry-3
You might want to see if the performance issue is on the Lasso side, or the
Apache side.

For Apache, you can log the total response time using %D (microseconds) or
%T (seconds):

http://httpd.apache.org/docs/current/mod/mod_log_config.html

Within Lasso, log the processing time for each page using a timer (or just
output it in a comment in the HTML and load the page yourself).  My bet is
the bulk of the page load time is in Lasso, not Apache, which means
optimizing your pages in Lasso, or perhaps ensuring databases are not
corrupted, the disk isn't nearly full, the internal logs are not packed
with years of rows, etc.


- Bil


On Mon, Mar 18, 2019 at 11:27 AM Jon Harris <[hidden email]> wrote:

> Hi
>
> We run Lasso 8 on a Windows VM under Apache 2.4.
>
> At peak times, (which is generally early evening) the site really slows
> down. We have been through our code adding "cache" tags everywhere:
>
> Cache(-Name="headerCache", -Expires=120) ;
> // the code....
> /cache ;
> (etc, etc)
>
> - and the overall performance has improved.
>
> However, it is still lumpy (we are seeing a number of 408 errors in the
> apache logs) We have tried adding additional processors and RAM to the VM -
> but this hasn't made any noticeable difference. As Lasso 8 is only 32-bit
> (although the connector is 64-bit) there doesn't seem any point allocating
> the server any more RAM, than the 8Gb already allocated to it.
>
> Looking at httpd.conf, I noticed there are a number of cache and proxy
> modules, most of which are commented out on our (default) installation.
>
> LoadModule cache_module modules/mod_cache.so LoadModule cache_disk_module
> modules/mod_cache_disk.so LoadModule cache_socache_module
> modules/mod_cache_socache.so LoadModule proxy_fcgi_module
> modules/mod_proxy_fcgi.so LoadModule proxy_hcheck_module
> modules/mod_proxy_hcheck.so LoadModule proxy_html_module
> modules/mod_proxy_html.so LoadModule proxy_http_module
> modules/mod_proxy_http.so LoadModule proxy_scgi_module
> modules/mod_proxy_scgi.so LoadModule proxy_uwsgi_module
> modules/mod_proxy_uwsgi.so LoadModule proxy_wstunnel_module
> modules/mod_proxy_wstunnel.so
>
> Do any of these modules make an difference to the performance of Lasso?
>
> Also, we have a dozen or so rewrite rules in our .htaccess.  Is it worth
> trying to ditch .htaccess and write those rules into the httpd.conf? My
> understanding is that for every file request Apache has to run the url
> through the .htaccess and as result it can run faster with without one.
>
> Beyond this, without major rewriting, would a front-end proxy server be
> the only viable solution.
>
> I'd be grateful for any useful advice.
>
> Regards
> Jon Harris
>
>
> #############################################################
>
> This message is sent to you because you are subscribed to
>   the mailing list Lasso [hidden email]
> Official list archives available at http://www.lassotalk.com
> To unsubscribe, E-mail to: <[hidden email]>
> Send administrative queries to  <[hidden email]>
>

#############################################################

This message is sent to you because you are subscribed to
  the mailing list Lasso [hidden email]
Official list archives available at http://www.lassotalk.com
To unsubscribe, E-mail to: <[hidden email]>
Send administrative queries to  <[hidden email]>
Reply | Threaded
Open this post in threaded view
|

RE: Lasso 8 on Apache

Jon Harris-2
Hi Bill

Thanks for the response.

I saved the "generated" html as a straight .html page and ran a user simulator on both the .html and .lasso files.  

It's not brilliant and I am sure there are better tools, but we are using the "Paessler Webserver Stress Tool 8" to simulate users against 4 or 5 busy urls. From what we found the .html files can cope with the load and the .lasso versions of the same pages struggle.

- So it does look like it is Lasso that's the bottleneck.

We have taken mySQL out of the loop by using the [cache] tags around the database calls, although for the most part we are using Lasso to write html - from the data in the CMS. Rather than use records loops directly on the page.

Google analytics is report about 500 "current" users - so it is quite busy at that time.

My question was about the affect of the Apache cache modules when running Lasso. Do you have any experience with them?

Regards
jon

-----Original Message-----
From: [hidden email] <[hidden email]> On Behalf Of Bil Corry
Sent: 18 March 2019 03:49 PM
To: [hidden email]
Subject: Re: Lasso 8 on Apache

You might want to see if the performance issue is on the Lasso side, or the Apache side.

For Apache, you can log the total response time using %D (microseconds) or %T (seconds):

http://httpd.apache.org/docs/current/mod/mod_log_config.html

Within Lasso, log the processing time for each page using a timer (or just output it in a comment in the HTML and load the page yourself).  My bet is the bulk of the page load time is in Lasso, not Apache, which means optimizing your pages in Lasso, or perhaps ensuring databases are not corrupted, the disk isn't nearly full, the internal logs are not packed with years of rows, etc.


- Bil


On Mon, Mar 18, 2019 at 11:27 AM Jon Harris <[hidden email]> wrote:

> Hi
>
> We run Lasso 8 on a Windows VM under Apache 2.4.
>
> At peak times, (which is generally early evening) the site really
> slows down. We have been through our code adding "cache" tags everywhere:
>
> Cache(-Name="headerCache", -Expires=120) ; // the code....
> /cache ;
> (etc, etc)
>
> - and the overall performance has improved.
>
> However, it is still lumpy (we are seeing a number of 408 errors in
> the apache logs) We have tried adding additional processors and RAM to
> the VM - but this hasn't made any noticeable difference. As Lasso 8 is
> only 32-bit (although the connector is 64-bit) there doesn't seem any
> point allocating the server any more RAM, than the 8Gb already allocated to it.
>
> Looking at httpd.conf, I noticed there are a number of cache and proxy
> modules, most of which are commented out on our (default) installation.
>
> LoadModule cache_module modules/mod_cache.so LoadModule
> cache_disk_module modules/mod_cache_disk.so LoadModule
> cache_socache_module modules/mod_cache_socache.so LoadModule
> proxy_fcgi_module modules/mod_proxy_fcgi.so LoadModule
> proxy_hcheck_module modules/mod_proxy_hcheck.so LoadModule
> proxy_html_module modules/mod_proxy_html.so LoadModule
> proxy_http_module modules/mod_proxy_http.so LoadModule
> proxy_scgi_module modules/mod_proxy_scgi.so LoadModule
> proxy_uwsgi_module modules/mod_proxy_uwsgi.so LoadModule
> proxy_wstunnel_module modules/mod_proxy_wstunnel.so
>
> Do any of these modules make an difference to the performance of Lasso?
>
> Also, we have a dozen or so rewrite rules in our .htaccess.  Is it
> worth trying to ditch .htaccess and write those rules into the
> httpd.conf? My understanding is that for every file request Apache has
> to run the url through the .htaccess and as result it can run faster with without one.
>
> Beyond this, without major rewriting, would a front-end proxy server
> be the only viable solution.
>
> I'd be grateful for any useful advice.
>
> Regards
> Jon Harris
>
>
> #############################################################
>
> This message is sent to you because you are subscribed to
>   the mailing list Lasso [hidden email] Official list
> archives available at http://www.lassotalk.com To unsubscribe, E-mail
> to: <[hidden email]>
> Send administrative queries to  <[hidden email]>
>

#############################################################

This message is sent to you because you are subscribed to
  the mailing list Lasso [hidden email] Official list archives available at http://www.lassotalk.com To unsubscribe, E-mail to: <[hidden email]>
Send administrative queries to  <[hidden email]>

#############################################################

This message is sent to you because you are subscribed to
  the mailing list Lasso [hidden email]
Official list archives available at http://www.lassotalk.com
To unsubscribe, E-mail to: <[hidden email]>
Send administrative queries to  <[hidden email]>
Reply | Threaded
Open this post in threaded view
|

Re: Lasso 8 on Apache

Bil Corry-3
I don't have any experience using the cache modules, sorry.  You might also
look into load balancing and run more than a single server.

- Bil


On Mon, Mar 18, 2019 at 12:55 PM Jon Harris <[hidden email]> wrote:

> Hi Bill
>
> Thanks for the response.
>
> I saved the "generated" html as a straight .html page and ran a user
> simulator on both the .html and .lasso files.
>
> It's not brilliant and I am sure there are better tools, but we are using
> the "Paessler Webserver Stress Tool 8" to simulate users against 4 or 5
> busy urls. From what we found the .html files can cope with the load and
> the .lasso versions of the same pages struggle.
>
> - So it does look like it is Lasso that's the bottleneck.
>
> We have taken mySQL out of the loop by using the [cache] tags around the
> database calls, although for the most part we are using Lasso to write html
> - from the data in the CMS. Rather than use records loops directly on the
> page.
>
> Google analytics is report about 500 "current" users - so it is quite busy
> at that time.
>
> My question was about the affect of the Apache cache modules when running
> Lasso. Do you have any experience with them?
>
> Regards
> jon
>
> -----Original Message-----
> From: [hidden email] <[hidden email]>
> On Behalf Of Bil Corry
> Sent: 18 March 2019 03:49 PM
> To: [hidden email]
> Subject: Re: Lasso 8 on Apache
>
> You might want to see if the performance issue is on the Lasso side, or
> the Apache side.
>
> For Apache, you can log the total response time using %D (microseconds) or
> %T (seconds):
>
> http://httpd.apache.org/docs/current/mod/mod_log_config.html
>
> Within Lasso, log the processing time for each page using a timer (or just
> output it in a comment in the HTML and load the page yourself).  My bet is
> the bulk of the page load time is in Lasso, not Apache, which means
> optimizing your pages in Lasso, or perhaps ensuring databases are not
> corrupted, the disk isn't nearly full, the internal logs are not packed
> with years of rows, etc.
>
>
> - Bil
>
>
> On Mon, Mar 18, 2019 at 11:27 AM Jon Harris <[hidden email]> wrote:
>
> > Hi
> >
> > We run Lasso 8 on a Windows VM under Apache 2.4.
> >
> > At peak times, (which is generally early evening) the site really
> > slows down. We have been through our code adding "cache" tags everywhere:
> >
> > Cache(-Name="headerCache", -Expires=120) ; // the code....
> > /cache ;
> > (etc, etc)
> >
> > - and the overall performance has improved.
> >
> > However, it is still lumpy (we are seeing a number of 408 errors in
> > the apache logs) We have tried adding additional processors and RAM to
> > the VM - but this hasn't made any noticeable difference. As Lasso 8 is
> > only 32-bit (although the connector is 64-bit) there doesn't seem any
> > point allocating the server any more RAM, than the 8Gb already allocated
> to it.
> >
> > Looking at httpd.conf, I noticed there are a number of cache and proxy
> > modules, most of which are commented out on our (default) installation.
> >
> > LoadModule cache_module modules/mod_cache.so LoadModule
> > cache_disk_module modules/mod_cache_disk.so LoadModule
> > cache_socache_module modules/mod_cache_socache.so LoadModule
> > proxy_fcgi_module modules/mod_proxy_fcgi.so LoadModule
> > proxy_hcheck_module modules/mod_proxy_hcheck.so LoadModule
> > proxy_html_module modules/mod_proxy_html.so LoadModule
> > proxy_http_module modules/mod_proxy_http.so LoadModule
> > proxy_scgi_module modules/mod_proxy_scgi.so LoadModule
> > proxy_uwsgi_module modules/mod_proxy_uwsgi.so LoadModule
> > proxy_wstunnel_module modules/mod_proxy_wstunnel.so
> >
> > Do any of these modules make an difference to the performance of Lasso?
> >
> > Also, we have a dozen or so rewrite rules in our .htaccess.  Is it
> > worth trying to ditch .htaccess and write those rules into the
> > httpd.conf? My understanding is that for every file request Apache has
> > to run the url through the .htaccess and as result it can run faster
> with without one.
> >
> > Beyond this, without major rewriting, would a front-end proxy server
> > be the only viable solution.
> >
> > I'd be grateful for any useful advice.
> >
> > Regards
> > Jon Harris
> >
> >
> > #############################################################
> >
> > This message is sent to you because you are subscribed to
> >   the mailing list Lasso [hidden email] Official list
> > archives available at http://www.lassotalk.com To unsubscribe, E-mail
> > to: <[hidden email]>
> > Send administrative queries to  <[hidden email]>
> >
>
> #############################################################
>
> This message is sent to you because you are subscribed to
>   the mailing list Lasso [hidden email] Official list archives
> available at http://www.lassotalk.com To unsubscribe, E-mail to: <
> [hidden email]>
> Send administrative queries to  <[hidden email]>
>
> #############################################################
>
> This message is sent to you because you are subscribed to
>   the mailing list Lasso [hidden email]
> Official list archives available at http://www.lassotalk.com
> To unsubscribe, E-mail to: <[hidden email]>
> Send administrative queries to  <[hidden email]>
>

#############################################################

This message is sent to you because you are subscribed to
  the mailing list Lasso [hidden email]
Official list archives available at http://www.lassotalk.com
To unsubscribe, E-mail to: <[hidden email]>
Send administrative queries to  <[hidden email]>
Reply | Threaded
Open this post in threaded view
|

Re: Lasso 8 on Apache

Mason Miller
In reply to this post by Jon Harris-2
How often does the data on the pages change?

You might get a lot of mileage using Apache’s caching instead of Lasso’s. You might even set up a different file extension for the pages you want cached and those that need to be dynamic every time.

Mason

Sent from my iPhone

> On Mar 18, 2019, at 12:55 PM, Jon Harris <[hidden email]> wrote:
>
> Hi Bill
>
> Thanks for the response.
>
> I saved the "generated" html as a straight .html page and ran a user simulator on both the .html and .lasso files.  
>
> It's not brilliant and I am sure there are better tools, but we are using the "Paessler Webserver Stress Tool 8" to simulate users against 4 or 5 busy urls. From what we found the .html files can cope with the load and the .lasso versions of the same pages struggle.
>
> - So it does look like it is Lasso that's the bottleneck.
>
> We have taken mySQL out of the loop by using the [cache] tags around the database calls, although for the most part we are using Lasso to write html - from the data in the CMS. Rather than use records loops directly on the page.
>
> Google analytics is report about 500 "current" users - so it is quite busy at that time.
>
> My question was about the affect of the Apache cache modules when running Lasso. Do you have any experience with them?
>
> Regards
> jon
>
> -----Original Message-----
> From: [hidden email] <[hidden email]> On Behalf Of Bil Corry
> Sent: 18 March 2019 03:49 PM
> To: [hidden email]
> Subject: Re: Lasso 8 on Apache
>
> You might want to see if the performance issue is on the Lasso side, or the Apache side.
>
> For Apache, you can log the total response time using %D (microseconds) or %T (seconds):
>
> http://httpd.apache.org/docs/current/mod/mod_log_config.html
>
> Within Lasso, log the processing time for each page using a timer (or just output it in a comment in the HTML and load the page yourself).  My bet is the bulk of the page load time is in Lasso, not Apache, which means optimizing your pages in Lasso, or perhaps ensuring databases are not corrupted, the disk isn't nearly full, the internal logs are not packed with years of rows, etc.
>
>
> - Bil
>
>
>> On Mon, Mar 18, 2019 at 11:27 AM Jon Harris <[hidden email]> wrote:
>>
>> Hi
>>
>> We run Lasso 8 on a Windows VM under Apache 2.4.
>>
>> At peak times, (which is generally early evening) the site really
>> slows down. We have been through our code adding "cache" tags everywhere:
>>
>> Cache(-Name="headerCache", -Expires=120) ; // the code....
>> /cache ;
>> (etc, etc)
>>
>> - and the overall performance has improved.
>>
>> However, it is still lumpy (we are seeing a number of 408 errors in
>> the apache logs) We have tried adding additional processors and RAM to
>> the VM - but this hasn't made any noticeable difference. As Lasso 8 is
>> only 32-bit (although the connector is 64-bit) there doesn't seem any
>> point allocating the server any more RAM, than the 8Gb already allocated to it.
>>
>> Looking at httpd.conf, I noticed there are a number of cache and proxy
>> modules, most of which are commented out on our (default) installation.
>>
>> LoadModule cache_module modules/mod_cache.so LoadModule
>> cache_disk_module modules/mod_cache_disk.so LoadModule
>> cache_socache_module modules/mod_cache_socache.so LoadModule
>> proxy_fcgi_module modules/mod_proxy_fcgi.so LoadModule
>> proxy_hcheck_module modules/mod_proxy_hcheck.so LoadModule
>> proxy_html_module modules/mod_proxy_html.so LoadModule
>> proxy_http_module modules/mod_proxy_http.so LoadModule
>> proxy_scgi_module modules/mod_proxy_scgi.so LoadModule
>> proxy_uwsgi_module modules/mod_proxy_uwsgi.so LoadModule
>> proxy_wstunnel_module modules/mod_proxy_wstunnel.so
>>
>> Do any of these modules make an difference to the performance of Lasso?
>>
>> Also, we have a dozen or so rewrite rules in our .htaccess.  Is it
>> worth trying to ditch .htaccess and write those rules into the
>> httpd.conf? My understanding is that for every file request Apache has
>> to run the url through the .htaccess and as result it can run faster with without one.
>>
>> Beyond this, without major rewriting, would a front-end proxy server
>> be the only viable solution.
>>
>> I'd be grateful for any useful advice.
>>
>> Regards
>> Jon Harris
>>
>>
>> #############################################################
>>
>> This message is sent to you because you are subscribed to
>>  the mailing list Lasso [hidden email] Official list
>> archives available at http://www.lassotalk.com To unsubscribe, E-mail
>> to: <[hidden email]>
>> Send administrative queries to  <[hidden email]>
>>
>
> #############################################################
>
> This message is sent to you because you are subscribed to
>  the mailing list Lasso [hidden email] Official list archives available at http://www.lassotalk.com To unsubscribe, E-mail to: <[hidden email]>
> Send administrative queries to  <[hidden email]>
>
> #############################################################
>
> This message is sent to you because you are subscribed to
>  the mailing list Lasso [hidden email]
> Official list archives available at http://www.lassotalk.com
> To unsubscribe, E-mail to: <[hidden email]>
> Send administrative queries to  <[hidden email]>



#############################################################

This message is sent to you because you are subscribed to
  the mailing list Lasso [hidden email]
Official list archives available at http://www.lassotalk.com
To unsubscribe, E-mail to: <[hidden email]>
Send administrative queries to  <[hidden email]>
Reply | Threaded
Open this post in threaded view
|

Re: Lasso 8 on Apache

Jussi Hirvi-2
In reply to this post by Jon Harris-2
Could you try it under Linux instead? Linux would anyway be easier to
maintain than Windows.

It would stand to reason that Apache would perform better on Linux, as
there is no obligatory GUI, and also because Apache was developed for
Unix/Linux first, and only then ported to Windows.

Lasso speed under Windows vs. Linux is another matter, and probably the
decisive one for you. I don't know about that. But it should not be a
very big effort to test it.

- Jussi

On 18.3.2019 17.27, Jon Harris wrote:
> We run Lasso 8 on a Windows VM under Apache 2.4.

#############################################################

This message is sent to you because you are subscribed to
  the mailing list Lasso [hidden email]
Official list archives available at http://www.lassotalk.com
To unsubscribe, E-mail to: <[hidden email]>
Send administrative queries to  <[hidden email]>
Reply | Threaded
Open this post in threaded view
|

RE: Lasso 8 on Apache

Jon Harris-2
In reply to this post by Mason Miller
Hi Mason

Thanks for responding to my post.

It's quite an interesting site -it is showing sports results. So we have 1000s of views of the same data in a relatively short period. Once the games are over, the data is fairly static. However, we want the maximum delay to be two minutes, so during games visitors can see a reasonably updated current score.  We have set the cache timeout to 120 seconds, meaning the average delay is one minute.

As I type this, I am thinking that it might be possible to set the cache expiry dynamically higher - after a game has finished.

How, do we  use Apache's caching instead of Lasso's. Do we remove the "cache" tags in our code and change something in the httpd.conf?

I was thinking of creating an NGINX server as a proxy front-end server, or using proxy hosting service such as cloudFront, but this is quite a major project and we are concerned about how we can control the cache times.

Regards
Jon Harris

-----Original Message-----
From: [hidden email] <[hidden email]> On Behalf Of Mason Miller
Sent: 18 March 2019 09:44 PM
To: [hidden email]
Subject: Re: Lasso 8 on Apache

How often does the data on the pages change?

You might get a lot of mileage using Apache’s caching instead of Lasso’s. You might even set up a different file extension for the pages you want cached and those that need to be dynamic every time.

Mason

Sent from my iPhone

> On Mar 18, 2019, at 12:55 PM, Jon Harris <[hidden email]> wrote:
>
> Hi Bill
>
> Thanks for the response.
>
> I saved the "generated" html as a straight .html page and ran a user simulator on both the .html and .lasso files.  
>
> It's not brilliant and I am sure there are better tools, but we are using the "Paessler Webserver Stress Tool 8" to simulate users against 4 or 5 busy urls. From what we found the .html files can cope with the load and the .lasso versions of the same pages struggle.
>
> - So it does look like it is Lasso that's the bottleneck.
>
> We have taken mySQL out of the loop by using the [cache] tags around the database calls, although for the most part we are using Lasso to write html - from the data in the CMS. Rather than use records loops directly on the page.
>
> Google analytics is report about 500 "current" users - so it is quite busy at that time.
>
> My question was about the affect of the Apache cache modules when running Lasso. Do you have any experience with them?
>
> Regards
> jon
>
> -----Original Message-----
> From: [hidden email]
> <[hidden email]> On Behalf Of Bil Corry
> Sent: 18 March 2019 03:49 PM
> To: [hidden email]
> Subject: Re: Lasso 8 on Apache
>
> You might want to see if the performance issue is on the Lasso side, or the Apache side.
>
> For Apache, you can log the total response time using %D (microseconds) or %T (seconds):
>
> http://httpd.apache.org/docs/current/mod/mod_log_config.html
>
> Within Lasso, log the processing time for each page using a timer (or just output it in a comment in the HTML and load the page yourself).  My bet is the bulk of the page load time is in Lasso, not Apache, which means optimizing your pages in Lasso, or perhaps ensuring databases are not corrupted, the disk isn't nearly full, the internal logs are not packed with years of rows, etc.
>
>
> - Bil
>
>
>> On Mon, Mar 18, 2019 at 11:27 AM Jon Harris <[hidden email]> wrote:
>>
>> Hi
>>
>> We run Lasso 8 on a Windows VM under Apache 2.4.
>>
>> At peak times, (which is generally early evening) the site really
>> slows down. We have been through our code adding "cache" tags everywhere:
>>
>> Cache(-Name="headerCache", -Expires=120) ; // the code....
>> /cache ;
>> (etc, etc)
>>
>> - and the overall performance has improved.
>>
>> However, it is still lumpy (we are seeing a number of 408 errors in
>> the apache logs) We have tried adding additional processors and RAM
>> to the VM - but this hasn't made any noticeable difference. As Lasso
>> 8 is only 32-bit (although the connector is 64-bit) there doesn't
>> seem any point allocating the server any more RAM, than the 8Gb already allocated to it.
>>
>> Looking at httpd.conf, I noticed there are a number of cache and
>> proxy modules, most of which are commented out on our (default) installation.
>>
>> LoadModule cache_module modules/mod_cache.so LoadModule
>> cache_disk_module modules/mod_cache_disk.so LoadModule
>> cache_socache_module modules/mod_cache_socache.so LoadModule
>> proxy_fcgi_module modules/mod_proxy_fcgi.so LoadModule
>> proxy_hcheck_module modules/mod_proxy_hcheck.so LoadModule
>> proxy_html_module modules/mod_proxy_html.so LoadModule
>> proxy_http_module modules/mod_proxy_http.so LoadModule
>> proxy_scgi_module modules/mod_proxy_scgi.so LoadModule
>> proxy_uwsgi_module modules/mod_proxy_uwsgi.so LoadModule
>> proxy_wstunnel_module modules/mod_proxy_wstunnel.so
>>
>> Do any of these modules make an difference to the performance of Lasso?
>>
>> Also, we have a dozen or so rewrite rules in our .htaccess.  Is it
>> worth trying to ditch .htaccess and write those rules into the
>> httpd.conf? My understanding is that for every file request Apache
>> has to run the url through the .htaccess and as result it can run faster with without one.
>>
>> Beyond this, without major rewriting, would a front-end proxy server
>> be the only viable solution.
>>
>> I'd be grateful for any useful advice.
>>
>> Regards
>> Jon Harris
>>
>>
>> #############################################################
>>
>> This message is sent to you because you are subscribed to  the
>> mailing list Lasso [hidden email] Official list archives
>> available at http://www.lassotalk.com To unsubscribe, E-mail
>> to: <[hidden email]>
>> Send administrative queries to  <[hidden email]>
>>
>
> #############################################################
>
> This message is sent to you because you are subscribed to  the mailing
> list Lasso [hidden email] Official list archives available
> at http://www.lassotalk.com To unsubscribe, E-mail to:
> <[hidden email]>
> Send administrative queries to  <[hidden email]>
>
> #############################################################
>
> This message is sent to you because you are subscribed to  the mailing
> list Lasso [hidden email] Official list archives available
> at http://www.lassotalk.com To unsubscribe, E-mail to:
> <[hidden email]>
> Send administrative queries to  <[hidden email]>



#############################################################

This message is sent to you because you are subscribed to
  the mailing list Lasso [hidden email] Official list archives available at http://www.lassotalk.com To unsubscribe, E-mail to: <[hidden email]>
Send administrative queries to  <[hidden email]>

#############################################################

This message is sent to you because you are subscribed to
  the mailing list Lasso [hidden email]
Official list archives available at http://www.lassotalk.com
To unsubscribe, E-mail to: <[hidden email]>
Send administrative queries to  <[hidden email]>
Reply | Threaded
Open this post in threaded view
|

RE: Lasso 8 on Apache

Jon Harris-2
In reply to this post by Jussi Hirvi-2
Hi Jussi

As I am sure you know, Windows isn't case-sensitive, but Linux is. There is quite a lot to go through to fix this, although we would do it - if we could guarantee this was definitely the solution.

This isn't just a Lasso issue, we would have to check every <img> and <script>, even the font files in the css files. Also, due to the limitations in lasso - we do a lot of os_process calls, so if someone uploads, for example a HEIC image file from an iPhone, we use a command line to convert it to a jpeg. (ImageMagick in Lasso is too old).

So switching to Linux won't be as simple as copying the files across and changing the DNS over.

Regards
Jon


-----Original Message-----
From: [hidden email] <[hidden email]> On Behalf Of Jussi Hirvi
Sent: 19 March 2019 05:56 AM
To: Lasso Talk <[hidden email]>
Subject: Re: Lasso 8 on Apache

Could you try it under Linux instead? Linux would anyway be easier to maintain than Windows.

It would stand to reason that Apache would perform better on Linux, as there is no obligatory GUI, and also because Apache was developed for Unix/Linux first, and only then ported to Windows.

Lasso speed under Windows vs. Linux is another matter, and probably the decisive one for you. I don't know about that. But it should not be a very big effort to test it.

- Jussi

On 18.3.2019 17.27, Jon Harris wrote:
> We run Lasso 8 on a Windows VM under Apache 2.4.

#############################################################

This message is sent to you because you are subscribed to
  the mailing list Lasso [hidden email] Official list archives available at http://www.lassotalk.com To unsubscribe, E-mail to: <[hidden email]>
Send administrative queries to  <[hidden email]>

#############################################################

This message is sent to you because you are subscribed to
  the mailing list Lasso [hidden email]
Official list archives available at http://www.lassotalk.com
To unsubscribe, E-mail to: <[hidden email]>
Send administrative queries to  <[hidden email]>
Reply | Threaded
Open this post in threaded view
|

Re: Lasso 8 on Apache

Mark Palmer
In reply to this post by Jon Harris-2
Jon,

Could you just load the part of the page that is changing using
XMLHttpRequest (XHR)? If it’s just a few lines of scores as text it
could potentially be a massively reduced request in terms of size and
complexity.

On 19 Mar 2019, at 10:09, Jon Harris wrote:

> It's quite an interesting site -it is showing sports results. So we
> have 1000s of views of the same data in a relatively short period.
> Once the games are over, the data is fairly static.

Regards

Mark Palmer
Pageworks Limited
E: [hidden email]
T: 01285 610035 and 01902 620500
W: www.pageworks.co.uk

#############################################################

This message is sent to you because you are subscribed to
  the mailing list Lasso [hidden email]
Official list archives available at http://www.lassotalk.com
To unsubscribe, E-mail to: <[hidden email]>
Send administrative queries to  <[hidden email]>
Reply | Threaded
Open this post in threaded view
|

RE: Lasso 8 on Apache

stevepiercy
In reply to this post by Jon Harris-2
Hi Jon,

When data is updated, you need to update the cache.  Why not do
that programmatically in Lasso instead of setting a low cache
expiration time?

https://reference.lassosoft.com/LassoReference.LassoApp?[Cache_Delete]
https://reference.lassosoft.com/LassoReference.LassoApp?[Cache_Empty]
https://reference.lassosoft.com/LassoReference.LassoApp?[Cache_Store]
https://reference.lassosoft.com/LassoReference.LassoApp?[Cache_Fetch]

You could add other caching mechanisms (web server, redis,
CloudFlare, etc.), but that adds complexity and layers of
caching that make it harder to debug.

--steve


On 3/19/19 at 10:09 AM, [hidden email] (Jon Harris) pronounced:

>Hi Mason
>
>Thanks for responding to my post.
>
>It's quite an interesting site -it is showing sports results.
>So we have 1000s of views of the same data in a relatively
>short period. Once the games are over, the data is fairly
>static. However, we want the maximum delay to be two minutes,
>so during games visitors can see a reasonably updated current
>score.  We have set the cache timeout to 120 seconds, meaning
>the average delay is one minute.
>As I type this, I am thinking that it might be possible to set
>the cache expiry dynamically higher - after a game has finished.
>How, do we  use Apache's caching instead of Lasso's. Do we
>remove the "cache" tags in our code and change something in the httpd.conf?
>
>I was thinking of creating an NGINX server as a proxy front-end
>server, or using proxy hosting service such as cloudFront, but
>this is quite a major project and we are concerned about how we
>can control the cache times.
>
>Regards
>Jon Harris
>
>-----Original Message-----
>From: [hidden email]
><[hidden email]> On Behalf Of Mason Miller
>Sent: 18 March 2019 09:44 PM
>To: [hidden email]
>Subject: Re: Lasso 8 on Apache
>
>How often does the data on the pages change?
>
>You might get a lot of mileage using Apache’s caching instead
>of Lasso’s. You might even set up a different file extension
>for the pages you want cached and those that need to be dynamic
>every time.
>Mason
>
>Sent from my iPhone
>
>>On Mar 18, 2019, at 12:55 PM, Jon Harris <[hidden email]> wrote:
>>
>>Hi Bill
>>
>>Thanks for the response.
>>
>>I saved the "generated" html as a straight .html page and ran a user simulator on both
>the .html and .lasso files.
>>
>>It's not brilliant and I am sure there are better tools, but we are using the "Paessler
>Webserver Stress Tool 8" to simulate users against 4 or 5 busy
>urls. From what we found the .html files can cope with the load
>and the .lasso versions of the same pages struggle.
>>
>>- So it does look like it is Lasso that's the bottleneck.
>>
>>We have taken mySQL out of the loop by using the [cache] tags around the database calls,
>although for the most part we are using Lasso to write html -
>from the data in the CMS. Rather than use records loops
>directly on the page.
>>
>>Google analytics is report about 500 "current" users - so it is quite busy at that time.
>>
>>My question was about the affect of the Apache cache modules when running Lasso. Do you
>have any experience with them?
>>
>>Regards
>>jon
>>
>>-----Original Message-----
>>From: [hidden email]
>><[hidden email]> On Behalf Of Bil Corry
>>Sent: 18 March 2019 03:49 PM
>>To: [hidden email]
>>Subject: Re: Lasso 8 on Apache
>>
>>You might want to see if the performance issue is on the Lasso side, or the Apache side.
>>
>>For Apache, you can log the total response time using %D (microseconds) or %T (seconds):
>>
>>http://httpd.apache.org/docs/current/mod/mod_log_config.html
>>
>>Within Lasso, log the processing time for each page using a timer (or just output it in a
>comment in the HTML and load the page yourself).  My bet is the
>bulk of the page load time is in Lasso, not Apache, which means
>optimizing your pages in Lasso, or perhaps ensuring databases
>are not corrupted, the disk isn't nearly full, the internal
>logs are not packed with years of rows, etc.
>>
>>
>>- Bil
>>
>>
>>> On Mon, Mar 18, 2019 at 11:27 AM Jon Harris <[hidden email]> wrote:
>>>   Hi
>>>   We run Lasso 8 on a Windows VM under Apache 2.4.
>>>   At peak times, (which is generally early evening) the site
>>>really  slows down. We have been through our code adding
>>>"cache" tags everywhere:
>>>   Cache(-Name="headerCache", -Expires=120) ; // the code....
>>> /cache ;
>>> (etc, etc)
>>>   - and the overall performance has improved.
>>>   However, it is still lumpy (we are seeing a number of 408
>>>errors in  the apache logs) We have tried adding additional
>>>processors and RAM  to the VM - but this hasn't made any
>>>noticeable difference. As Lasso  8 is only 32-bit (although
>>>the connector is 64-bit) there doesn't  seem any point
>>>allocating the server any more RAM, than the 8Gb already
>>>allocated to it.
>>>   Looking at httpd.conf, I noticed there are a number of
>>>cache and  proxy modules, most of which are commented out on
>>>our (default) installation.
>>>   LoadModule cache_module modules/mod_cache.so LoadModule  
>>>cache_disk_module modules/mod_cache_disk.so LoadModule  
>>>cache_socache_module modules/mod_cache_socache.so LoadModule  
>>>proxy_fcgi_module modules/mod_proxy_fcgi.so LoadModule  
>>>proxy_hcheck_module modules/mod_proxy_hcheck.so LoadModule  
>>>proxy_html_module modules/mod_proxy_html.so LoadModule  
>>>proxy_http_module modules/mod_proxy_http.so LoadModule  
>>>proxy_scgi_module modules/mod_proxy_scgi.so LoadModule  
>>>proxy_uwsgi_module modules/mod_proxy_uwsgi.so LoadModule  
>>>proxy_wstunnel_module modules/mod_proxy_wstunnel.so
>>>   Do any of these modules make an difference to the
>>>performance of Lasso?
>>>   Also, we have a dozen or so rewrite rules in our
>>>.htaccess.  Is it  worth trying to ditch .htaccess and write
>>>those rules into the  httpd.conf? My understanding is that
>>>for every file request Apache  has to run the url through the
>>>.htaccess and as result it can run faster with without
>one.
>>>   Beyond this, without major rewriting, would a front-end
>>>proxy server  be the only viable solution.
>>>   I'd be grateful for any useful advice.
>>>   Regards
>>> Jon Harris
>>>    #############################################################
>>>   This message is sent to you because you are subscribed to  
>>>the  mailing list Lasso [hidden email] Official
>>>list archives  available at http://www.lassotalk.com To
>>>unsubscribe, E-mail
>>> to: <[hidden email]>
>>> Send administrative queries to  <[hidden email]>
>>>
>>
>>#############################################################
>>
>>This message is sent to you because you are subscribed to  the
>>mailing list Lasso [hidden email] Official list
>>archives available at http://www.lassotalk.com To unsubscribe,
>>E-mail to: <[hidden email]>
>>Send administrative queries to  <[hidden email]>
>>
>>#############################################################
>>
>>This message is sent to you because you are subscribed to  the
>>mailing list Lasso [hidden email] Official list
>>archives available at http://www.lassotalk.com To unsubscribe,
>>E-mail to: <[hidden email]>
>>Send administrative queries to  <[hidden email]>
>
>
>
>#############################################################
>
>This message is sent to you because you are subscribed to
>the mailing list Lasso [hidden email] Official list
>archives available at http://www.lassotalk.com To unsubscribe,
>E-mail to: <[hidden email]>
>Send administrative queries to  <[hidden email]>
>
>#############################################################
>
>This message is sent to you because you are subscribed to
>the mailing list Lasso [hidden email]
>Official list archives available at http://www.lassotalk.com
>To unsubscribe, E-mail to: <[hidden email]>
>Send administrative queries to  <[hidden email]>

-- -- -- -- -- -- -- -- -- -- -- -- -- -- -- -- -- -- -- -- -- --
Steve Piercy              Website Builder              Eugene, OR
<[hidden email]>               <http://www.stevepiercy.com/>


#############################################################

This message is sent to you because you are subscribed to
  the mailing list Lasso [hidden email]
Official list archives available at http://www.lassotalk.com
To unsubscribe, E-mail to: <[hidden email]>
Send administrative queries to  <[hidden email]>
Reply | Threaded
Open this post in threaded view
|

RE: Lasso 8 on Apache

Jon Harris-2
Hi Steve

That is interesting idea - I hadn't thought of that as a solution. I could zap the cache in the CMS and set a longer cache on the public side.

I have had a few problems with cache tags. For example, if we cache the headers (which are fairly static) the page title is the one from the first visitor to reload the cache. So I have created separate cache headers on each page - which is less than optimal.

The other problem is the page parameters. I have had to do things like:

cache(-name="resultsCache" + action_param('param1')+ action_param('param2'),-expires=120) ;

In the cms, I will see if I can loop these discreet params and delete all the potential cache names.

(Just to point out -  we don't use actually use "action_param" - it's actually a sanitised VAR()  )

I am looking at creating a "nginx.ourdomain.com" as a front-end proxy server - to see how it will behave and how well we can control the cache.

Ultimately, I think the issue is down to the Lasso architecture.  Apache and the Lasso connector are 64-bit, but the lasso service is 32-bit, which is the likely performance limiter. In which case a front-end solution feels like it will give us some future-proofing for traffic spikes and growth.

Regards
Jon Harris



-----Original Message-----
From: [hidden email] <[hidden email]> On Behalf Of Steve Piercy - Website Builder
Sent: 19 March 2019 06:13 PM
To: [hidden email]
Subject: RE: Lasso 8 on Apache

Hi Jon,

When data is updated, you need to update the cache.  Why not do that programmatically in Lasso instead of setting a low cache expiration time?

https://reference.lassosoft.com/LassoReference.LassoApp?[Cache_Delete]
https://reference.lassosoft.com/LassoReference.LassoApp?[Cache_Empty]
https://reference.lassosoft.com/LassoReference.LassoApp?[Cache_Store]
https://reference.lassosoft.com/LassoReference.LassoApp?[Cache_Fetch]

You could add other caching mechanisms (web server, redis, CloudFlare, etc.), but that adds complexity and layers of caching that make it harder to debug.

--steve


On 3/19/19 at 10:09 AM, [hidden email] (Jon Harris) pronounced:

>Hi Mason
>
>Thanks for responding to my post.
>
>It's quite an interesting site -it is showing sports results.
>So we have 1000s of views of the same data in a relatively short
>period. Once the games are over, the data is fairly static. However, we
>want the maximum delay to be two minutes, so during games visitors can
>see a reasonably updated current score.  We have set the cache timeout
>to 120 seconds, meaning the average delay is one minute.
>As I type this, I am thinking that it might be possible to set the
>cache expiry dynamically higher - after a game has finished.
>How, do we  use Apache's caching instead of Lasso's. Do we remove the
>"cache" tags in our code and change something in the httpd.conf?
>
>I was thinking of creating an NGINX server as a proxy front-end server,
>or using proxy hosting service such as cloudFront, but this is quite a
>major project and we are concerned about how we can control the cache
>times.
>
>Regards
>Jon Harris
>
>-----Original Message-----
>From: [hidden email]
><[hidden email]> On Behalf Of Mason Miller
>Sent: 18 March 2019 09:44 PM
>To: [hidden email]
>Subject: Re: Lasso 8 on Apache
>
>How often does the data on the pages change?
>
>You might get a lot of mileage using Apache’s caching instead of
>Lasso’s. You might even set up a different file extension for the pages
>you want cached and those that need to be dynamic every time.
>Mason
>
>Sent from my iPhone
>
>>On Mar 18, 2019, at 12:55 PM, Jon Harris <[hidden email]> wrote:
>>
>>Hi Bill
>>
>>Thanks for the response.
>>
>>I saved the "generated" html as a straight .html page and ran a user
>>simulator on both
>the .html and .lasso files.
>>
>>It's not brilliant and I am sure there are better tools, but we are
>>using the "Paessler
>Webserver Stress Tool 8" to simulate users against 4 or 5 busy urls.
>From what we found the .html files can cope with the load and the
>.lasso versions of the same pages struggle.
>>
>>- So it does look like it is Lasso that's the bottleneck.
>>
>>We have taken mySQL out of the loop by using the [cache] tags around
>>the database calls,
>although for the most part we are using Lasso to write html - from the
>data in the CMS. Rather than use records loops directly on the page.
>>
>>Google analytics is report about 500 "current" users - so it is quite busy at that time.
>>
>>My question was about the affect of the Apache cache modules when
>>running Lasso. Do you
>have any experience with them?
>>
>>Regards
>>jon
>>
>>-----Original Message-----
>>From: [hidden email]
>><[hidden email]> On Behalf Of Bil Corry
>>Sent: 18 March 2019 03:49 PM
>>To: [hidden email]
>>Subject: Re: Lasso 8 on Apache
>>
>>You might want to see if the performance issue is on the Lasso side, or the Apache side.
>>
>>For Apache, you can log the total response time using %D (microseconds) or %T (seconds):
>>
>>http://httpd.apache.org/docs/current/mod/mod_log_config.html
>>
>>Within Lasso, log the processing time for each page using a timer (or
>>just output it in a
>comment in the HTML and load the page yourself).  My bet is the bulk of
>the page load time is in Lasso, not Apache, which means optimizing your
>pages in Lasso, or perhaps ensuring databases are not corrupted, the
>disk isn't nearly full, the internal logs are not packed with years of
>rows, etc.
>>
>>
>>- Bil
>>
>>
>>> On Mon, Mar 18, 2019 at 11:27 AM Jon Harris <[hidden email]> wrote:
>>>   Hi
>>>   We run Lasso 8 on a Windows VM under Apache 2.4.
>>>   At peak times, (which is generally early evening) the site really  
>>>slows down. We have been through our code adding "cache" tags
>>>everywhere:
>>>   Cache(-Name="headerCache", -Expires=120) ; // the code....
>>> /cache ;
>>> (etc, etc)
>>>   - and the overall performance has improved.
>>>   However, it is still lumpy (we are seeing a number of 408 errors
>>>in  the apache logs) We have tried adding additional processors and
>>>RAM  to the VM - but this hasn't made any noticeable difference. As
>>>Lasso  8 is only 32-bit (although the connector is 64-bit) there
>>>doesn't  seem any point allocating the server any more RAM, than the
>>>8Gb already allocated to it.
>>>   Looking at httpd.conf, I noticed there are a number of cache and  
>>>proxy modules, most of which are commented out on our (default)
>>>installation.
>>>   LoadModule cache_module modules/mod_cache.so LoadModule
>>>cache_disk_module modules/mod_cache_disk.so LoadModule
>>>cache_socache_module modules/mod_cache_socache.so LoadModule
>>>proxy_fcgi_module modules/mod_proxy_fcgi.so LoadModule
>>>proxy_hcheck_module modules/mod_proxy_hcheck.so LoadModule
>>>proxy_html_module modules/mod_proxy_html.so LoadModule
>>>proxy_http_module modules/mod_proxy_http.so LoadModule
>>>proxy_scgi_module modules/mod_proxy_scgi.so LoadModule
>>>proxy_uwsgi_module modules/mod_proxy_uwsgi.so LoadModule
>>>proxy_wstunnel_module modules/mod_proxy_wstunnel.so
>>>   Do any of these modules make an difference to the performance of
>>>Lasso?
>>>   Also, we have a dozen or so rewrite rules in our .htaccess.  Is it  
>>>worth trying to ditch .htaccess and write those rules into the  
>>>httpd.conf? My understanding is that for every file request Apache  
>>>has to run the url through the .htaccess and as result it can run
>>>faster with without
>one.
>>>   Beyond this, without major rewriting, would a front-end proxy
>>>server  be the only viable solution.
>>>   I'd be grateful for any useful advice.
>>>   Regards
>>> Jon Harris
>>>    #############################################################
>>>   This message is sent to you because you are subscribed to the  
>>>mailing list Lasso [hidden email] Official list archives  
>>>available at http://www.lassotalk.com To unsubscribe, E-mail
>>> to: <[hidden email]>
>>> Send administrative queries to  <[hidden email]>
>>>
>>
>>#############################################################
>>
>>This message is sent to you because you are subscribed to  the mailing
>>list Lasso [hidden email] Official list archives available
>>at http://www.lassotalk.com To unsubscribe, E-mail to:
>><[hidden email]>
>>Send administrative queries to  <[hidden email]>
>>
>>#############################################################
>>
>>This message is sent to you because you are subscribed to  the mailing
>>list Lasso [hidden email] Official list archives available
>>at http://www.lassotalk.com To unsubscribe, E-mail to:
>><[hidden email]>
>>Send administrative queries to  <[hidden email]>
>
>
>
>#############################################################
>
>This message is sent to you because you are subscribed to the mailing
>list Lasso [hidden email] Official list archives available
>at http://www.lassotalk.com To unsubscribe, E-mail to:
><[hidden email]>
>Send administrative queries to  <[hidden email]>
>
>#############################################################
>
>This message is sent to you because you are subscribed to the mailing
>list Lasso [hidden email] Official list archives available
>at http://www.lassotalk.com To unsubscribe, E-mail to:
><[hidden email]>
>Send administrative queries to  <[hidden email]>

-- -- -- -- -- -- -- -- -- -- -- -- -- -- -- -- -- -- -- -- -- --
Steve Piercy              Website Builder              Eugene, OR
<[hidden email]>               <http://www.stevepiercy.com/>


#############################################################

This message is sent to you because you are subscribed to
  the mailing list Lasso [hidden email] Official list archives available at http://www.lassotalk.com To unsubscribe, E-mail to: <[hidden email]>
Send administrative queries to  <[hidden email]>

#############################################################

This message is sent to you because you are subscribed to
  the mailing list Lasso [hidden email]
Official list archives available at http://www.lassotalk.com
To unsubscribe, E-mail to: <[hidden email]>
Send administrative queries to  <[hidden email]>
Reply | Threaded
Open this post in threaded view
|

RE: Lasso 8 on Apache

stevepiercy
Have you enabled mysql slow query log and analyzed results?
https://dev.mysql.com/doc/refman/8.0/en/slow-query-log.html

Have you enabled a Lasso timer, logging timing, and analyzed results?
http://www.lassosoft.com/tagswap/detail/timer

Before you blame the architecture (where's your data to support
your claim?) or pursue other remedies, I would look for the
usual suspects: shit code and shit queries.  Everyone's code
smells.  My code does not smell like roses, and I'm sure my
legacy Lasso 8 apps would fall over under load.

--steve


On 3/21/19 at 9:39 AM, [hidden email] (Jon Harris) pronounced:

>Hi Steve
>
>That is interesting idea - I hadn't thought of that as a
>solution. I could zap the cache in the CMS and set a longer
>cache on the public side.
>
>I have had a few problems with cache tags. For example, if we
>cache the headers (which are fairly static) the page title is
>the one from the first visitor to reload the cache. So I have
>created separate cache headers on each page - which is less
>than optimal.
>
>The other problem is the page parameters. I have had to do things like:
>
>cache(-name="resultsCache" + action_param('param1')+ action_param('param2'),-expires=120) ;
>
>In the cms, I will see if I can loop these discreet params and
>delete all the potential cache names.
>
>(Just to point out -  we don't use actually use "action_param"
>- it's actually a sanitised VAR()  )
>
>I am looking at creating a "nginx.ourdomain.com" as a front-end
>proxy server - to see how it will behave and how well we can
>control the cache.
>
>Ultimately, I think the issue is down to the Lasso
>architecture.  Apache and the Lasso connector are 64-bit, but
>the lasso service is 32-bit, which is the likely performance
>limiter. In which case a front-end solution feels like it will
>give us some future-proofing for traffic spikes and growth.
>
>Regards
>Jon Harris
>
>
>
>-----Original Message-----
>From: [hidden email]
><[hidden email]> On Behalf Of Steve Piercy -
>Website Builder
>Sent: 19 March 2019 06:13 PM
>To: [hidden email]
>Subject: RE: Lasso 8 on Apache
>
>Hi Jon,
>
>When data is updated, you need to update the cache.  Why not do
>that programmatically in Lasso instead of setting a low cache
>expiration time?
>
>https://reference.lassosoft.com/LassoReference.LassoApp?[Cache_Delete]
>https://reference.lassosoft.com/LassoReference.LassoApp?[Cache_Empty]
>https://reference.lassosoft.com/LassoReference.LassoApp?[Cache_Store]
>https://reference.lassosoft.com/LassoReference.LassoApp?[Cache_Fetch]
>
>You could add other caching mechanisms (web server, redis,
>CloudFlare, etc.), but that adds complexity and layers of
>caching that make it harder to debug.
>
>--steve
>
>
>On 3/19/19 at 10:09 AM, [hidden email] (Jon Harris) pronounced:
>
>>Hi Mason
>>
>>Thanks for responding to my post.
>>
>>It's quite an interesting site -it is showing sports results.
>>So we have 1000s of views of the same data in a relatively
>>short period. Once the games are over, the data is fairly
>>static. However, we want the maximum delay to be two minutes,
>>so during games visitors can see a reasonably updated current
>>score.  We have set the cache timeout to 120 seconds, meaning
>>the average delay is one minute.
>>As I type this, I am thinking that it might be possible to set
>>the cache expiry dynamically higher - after a game has finished.
>>How, do we  use Apache's caching instead of Lasso's. Do we
>>remove the "cache" tags in our code and change something in
>>the httpd.conf?
>>
>>I was thinking of creating an NGINX server as a proxy
>>front-end server, or using proxy hosting service such as
>>cloudFront, but this is quite a major project and we are
>>concerned about how we can control the cache times.
>>
>>Regards
>>Jon Harris
>>
>>-----Original Message-----
>>From: [hidden email]
>><[hidden email]> On Behalf Of Mason Miller
>>Sent: 18 March 2019 09:44 PM
>>To: [hidden email]
>>Subject: Re: Lasso 8 on Apache
>>
>>How often does the data on the pages change?
>>
>>You might get a lot of mileage using Apache’s caching
>>instead of Lasso’s. You might even set up a different file
>>extension for the pages you want cached and those that need to
>>be dynamic every time.
>>Mason
>>
>>Sent from my iPhone
>>
>>>On Mar 18, 2019, at 12:55 PM, Jon Harris <[hidden email]> wrote:
>>>
>>>Hi Bill
>>>
>>>Thanks for the response.
>>>
>>>I saved the "generated" html as a straight .html page and ran
>>>a user simulator on both
>>the .html and .lasso files.
>>>
>>>It's not brilliant and I am sure there are better tools, but
>>>we are using the "Paessler
>>Webserver Stress Tool 8" to simulate users against 4 or 5 busy
>>urls. From what we found the .html files can cope with the
>>load and the .lasso versions of the same pages struggle.
>>>
>>>- So it does look like it is Lasso that's the bottleneck.
>>>
>>>We have taken mySQL out of the loop by using the [cache] tags
>>>around the database calls,
>>although for the most part we are using Lasso to write html -
>>from the data in the CMS. Rather than use records loops
>>directly on the page.
>>>
>>>Google analytics is report about 500 "current" users - so it is quite busy at that time.
>>>
>>>My question was about the affect of the Apache cache modules
>>>when running Lasso. Do you
>>have any experience with them?
>>>
>>>Regards
>>>jon
>>>
>>>-----Original Message-----
>>>From: [hidden email]
>>><[hidden email]> On Behalf Of Bil Corry
>>>Sent: 18 March 2019 03:49 PM
>>>To: [hidden email]
>>>Subject: Re: Lasso 8 on Apache
>>>
>>>You might want to see if the performance issue is on the Lasso side, or the Apache side.
>>>
>>>For Apache, you can log the total response time using %D (microseconds) or %T (seconds):
>>>
>>>http://httpd.apache.org/docs/current/mod/mod_log_config.html
>>>
>>>Within Lasso, log the processing time for each page using a
>>>timer (or just output it in a
>>comment in the HTML and load the page yourself).  My bet is
>>the bulk of the page load time is in Lasso, not Apache, which
>>means optimizing your pages in Lasso, or perhaps ensuring
>>databases are not corrupted, the disk isn't nearly full, the
>>internal logs are not packed with years of rows, etc.
>>>
>>>
>>>- Bil
>>>
>>>
>>>> On Mon, Mar 18, 2019 at 11:27 AM Jon Harris <[hidden email]> wrote:
>>>>   Hi
>>>>   We run Lasso 8 on a Windows VM under Apache 2.4.
>>>>    At peak times, (which is generally early evening) the
>>>>site really  slows down. We have been through our code
>>>>adding "cache" tags everywhere:
>>>>   Cache(-Name="headerCache", -Expires=120) ; // the code....
>>>> /cache ;
>>>> (etc, etc)
>>>>   - and the overall performance has improved.
>>>>    However, it is still lumpy (we are seeing a number of
>>>>408 errors in  the apache logs) We have tried adding
>>>>additional processors and RAM  to the VM - but this hasn't
>>>>made any noticeable difference. As Lasso  8 is only 32-bit
>>>>(although the connector is 64-bit) there doesn't  seem any
>>>>point allocating the server any more RAM, than the 8Gb
>>>>already allocated to it.
>>>>    Looking at httpd.conf, I noticed there are a number of
>>>>cache and  proxy modules, most of which are commented out on
>>>>our (default) installation.
>>>>    LoadModule cache_module modules/mod_cache.so LoadModule
>>>>cache_disk_module modules/mod_cache_disk.so LoadModule
>>>>cache_socache_module modules/mod_cache_socache.so LoadModule
>>>>proxy_fcgi_module modules/mod_proxy_fcgi.so LoadModule
>>>>proxy_hcheck_module modules/mod_proxy_hcheck.so LoadModule
>>>>proxy_html_module modules/mod_proxy_html.so LoadModule
>>>>proxy_http_module modules/mod_proxy_http.so LoadModule
>>>>proxy_scgi_module modules/mod_proxy_scgi.so LoadModule
>>>>proxy_uwsgi_module modules/mod_proxy_uwsgi.so LoadModule
>>>>proxy_wstunnel_module modules/mod_proxy_wstunnel.so
>>>>    Do any of these modules make an difference to the
>>>>performance of Lasso?
>>>>    Also, we have a dozen or so rewrite rules in our
>>>>.htaccess.  Is it  worth trying to ditch .htaccess and write
>>>>those rules into the  httpd.conf? My understanding is that
>>>>for every file request Apache  has to run the url through
>>>>the .htaccess and as result it can run faster with without
>>one.
>>>>    Beyond this, without major rewriting, would a front-end
>>>>proxy server  be the only viable solution.
>>>>   I'd be grateful for any useful advice.
>>>>   Regards
>>>> Jon Harris
>>>>    #############################################################
>>>>    This message is sent to you because you are subscribed
>>>>to the  mailing list Lasso [hidden email]
>>>>Official list archives  available at
>>>>http://www.lassotalk.com To unsubscribe, E-mail
>>>> to: <[hidden email]>
>>>> Send administrative queries to  <[hidden email]>
>>>>
>>>
>>>#############################################################
>>>
>>>This message is sent to you because you are subscribed to  
>>>the mailing list Lasso [hidden email] Official
>>>list archives available at http://www.lassotalk.com To
>>>unsubscribe, E-mail to: <[hidden email]>
>>>Send administrative queries to  <[hidden email]>
>>>
>>>#############################################################
>>>
>>>This message is sent to you because you are subscribed to  
>>>the mailing list Lasso [hidden email] Official
>>>list archives available at http://www.lassotalk.com To
>>>unsubscribe, E-mail to: <[hidden email]>
>>>Send administrative queries to  <[hidden email]>
>>
>>
>>
>>#############################################################
>>
>>This message is sent to you because you are subscribed to the
>>mailing list Lasso [hidden email] Official list
>>archives available at http://www.lassotalk.com To unsubscribe,
>>E-mail to: <[hidden email]>
>>Send administrative queries to  <[hidden email]>
>>
>>#############################################################
>>
>>This message is sent to you because you are subscribed to the
>>mailing list Lasso [hidden email] Official list
>>archives available at http://www.lassotalk.com To unsubscribe,
>>E-mail to: <[hidden email]>
>>Send administrative queries to  <[hidden email]>
>
>-- -- -- -- -- -- -- -- -- -- -- -- -- -- -- -- -- -- -- -- -- --
>Steve Piercy              Website Builder              Eugene, OR
><[hidden email]>               <http://www.stevepiercy.com/>
>
>
>#############################################################
>
>This message is sent to you because you are subscribed to
>the mailing list Lasso [hidden email] Official list
>archives available at http://www.lassotalk.com To unsubscribe,
>E-mail to: <[hidden email]>
>Send administrative queries to  <[hidden email]>
>
>#############################################################
>
>This message is sent to you because you are subscribed to
>the mailing list Lasso [hidden email]
>Official list archives available at http://www.lassotalk.com
>To unsubscribe, E-mail to: <[hidden email]>
>Send administrative queries to  <[hidden email]>

-- -- -- -- -- -- -- -- -- -- -- -- -- -- -- -- -- -- -- -- -- --
Steve Piercy              Website Builder              Eugene, OR
<[hidden email]>               <http://www.stevepiercy.com/>


#############################################################

This message is sent to you because you are subscribed to
  the mailing list Lasso [hidden email]
Official list archives available at http://www.lassotalk.com
To unsubscribe, E-mail to: <[hidden email]>
Send administrative queries to  <[hidden email]>
Reply | Threaded
Open this post in threaded view
|

RE: Lasso 8 on Apache

Jon Harris-2
Hi Steve

Thanks for responding to my posting.Most of our CMS updates, where possible, write html which we [include] when we assemble the page - so probably not a query issue.

Yes, I have used Lasso timings. The [cache] tag has made the biggest difference to how fast lasso is. As I said, I saved the generated html as a separate html file and stress tested that, all my testing told me that as a flat-HTML site - it would cope fine.

Incidentally, as a test we moved to Google cloud-based hosted MySQL and it was around 2.5 times slower than our own cloud. (We have our own cloud servers, all our VMs talk to MySQL via a separate MPLS network. We have our own fibre-channel link to the EMC SAN.

I am sure our code could be optimised, as it has grown organically over a number of years. If we were starting from scratch now, it would almost certainly be quite different.

Every app will fail under load. Its' not possible to design computer systems to cope with an infinite amount of traffic.  If you have highly optimised code, using container objects. eventually, when you have filled a data centre with Kubernetes - your app will crash.

In fact our solution isn't failing too badly it's just taking around a minute for the home page, plus some critical pages to load when the server gets very busy. I'm sure if Lasso 8 was recompiled as a 64bit process I am pretty sure it would copy much better, but I am not blaming the technology - it is what it is.

I have decided to move to Nginx to proxy lasso requests, my early testing looks quite promising. I am just trying to get Nginx to serve the static content itself, as (this is not a Lasso issue) people are saying Nginx can serve static content much faster than Apache.

Regards
jon

-----Original Message-----
From: [hidden email] <[hidden email]> On Behalf Of Steve Piercy - Website Builder
Sent: 21 March 2019 01:19 PM
To: [hidden email]
Subject: RE: Lasso 8 on Apache

Have you enabled mysql slow query log and analyzed results?
https://dev.mysql.com/doc/refman/8.0/en/slow-query-log.html

Have you enabled a Lasso timer, logging timing, and analyzed results?
http://www.lassosoft.com/tagswap/detail/timer

Before you blame the architecture (where's your data to support your claim?) or pursue other remedies, I would look for the usual suspects: shit code and shit queries.  Everyone's code smells.  My code does not smell like roses, and I'm sure my legacy Lasso 8 apps would fall over under load.

--steve


On 3/21/19 at 9:39 AM, [hidden email] (Jon Harris) pronounced:

>Hi Steve
>
>That is interesting idea - I hadn't thought of that as a solution. I
>could zap the cache in the CMS and set a longer cache on the public
>side.
>
>I have had a few problems with cache tags. For example, if we cache the
>headers (which are fairly static) the page title is the one from the
>first visitor to reload the cache. So I have created separate cache
>headers on each page - which is less than optimal.
>
>The other problem is the page parameters. I have had to do things like:
>
>cache(-name="resultsCache" + action_param('param1')+
>action_param('param2'),-expires=120) ;
>
>In the cms, I will see if I can loop these discreet params and delete
>all the potential cache names.
>
>(Just to point out -  we don't use actually use "action_param"
>- it's actually a sanitised VAR()  )
>
>I am looking at creating a "nginx.ourdomain.com" as a front-end proxy
>server - to see how it will behave and how well we can control the
>cache.
>
>Ultimately, I think the issue is down to the Lasso architecture.  
>Apache and the Lasso connector are 64-bit, but the lasso service is
>32-bit, which is the likely performance limiter. In which case a
>front-end solution feels like it will give us some future-proofing for
>traffic spikes and growth.
>
>Regards
>Jon Harris
>
>
>
>-----Original Message-----
>From: [hidden email]
><[hidden email]> On Behalf Of Steve Piercy - Website
>Builder
>Sent: 19 March 2019 06:13 PM
>To: [hidden email]
>Subject: RE: Lasso 8 on Apache
>
>Hi Jon,
>
>When data is updated, you need to update the cache.  Why not do that
>programmatically in Lasso instead of setting a low cache expiration
>time?
>
>https://reference.lassosoft.com/LassoReference.LassoApp?[Cache_Delete]
>https://reference.lassosoft.com/LassoReference.LassoApp?[Cache_Empty]
>https://reference.lassosoft.com/LassoReference.LassoApp?[Cache_Store]
>https://reference.lassosoft.com/LassoReference.LassoApp?[Cache_Fetch]
>
>You could add other caching mechanisms (web server, redis, CloudFlare,
>etc.), but that adds complexity and layers of caching that make it
>harder to debug.
>
>--steve
>
>
>On 3/19/19 at 10:09 AM, [hidden email] (Jon Harris) pronounced:
>
>>Hi Mason
>>
>>Thanks for responding to my post.
>>
>>It's quite an interesting site -it is showing sports results.
>>So we have 1000s of views of the same data in a relatively short
>>period. Once the games are over, the data is fairly static. However,
>>we want the maximum delay to be two minutes, so during games visitors
>>can see a reasonably updated current score.  We have set the cache
>>timeout to 120 seconds, meaning the average delay is one minute.
>>As I type this, I am thinking that it might be possible to set the
>>cache expiry dynamically higher - after a game has finished.
>>How, do we  use Apache's caching instead of Lasso's. Do we remove the
>>"cache" tags in our code and change something in the httpd.conf?
>>
>>I was thinking of creating an NGINX server as a proxy front-end
>>server, or using proxy hosting service such as cloudFront, but this is
>>quite a major project and we are concerned about how we can control
>>the cache times.
>>
>>Regards
>>Jon Harris
>>
>>-----Original Message-----
>>From: [hidden email]
>><[hidden email]> On Behalf Of Mason Miller
>>Sent: 18 March 2019 09:44 PM
>>To: [hidden email]
>>Subject: Re: Lasso 8 on Apache
>>
>>How often does the data on the pages change?
>>
>>You might get a lot of mileage using Apache’s caching instead of
>>Lasso’s. You might even set up a different file extension for the
>>pages you want cached and those that need to be dynamic every time.
>>Mason
>>
>>Sent from my iPhone
>>
>>>On Mar 18, 2019, at 12:55 PM, Jon Harris <[hidden email]> wrote:
>>>
>>>Hi Bill
>>>
>>>Thanks for the response.
>>>
>>>I saved the "generated" html as a straight .html page and ran a user
>>>simulator on both
>>the .html and .lasso files.
>>>
>>>It's not brilliant and I am sure there are better tools, but we are
>>>using the "Paessler
>>Webserver Stress Tool 8" to simulate users against 4 or 5 busy urls.
>>From what we found the .html files can cope with the load and the
>>.lasso versions of the same pages struggle.
>>>
>>>- So it does look like it is Lasso that's the bottleneck.
>>>
>>>We have taken mySQL out of the loop by using the [cache] tags around
>>>the database calls,
>>although for the most part we are using Lasso to write html - from the
>>data in the CMS. Rather than use records loops directly on the page.
>>>
>>>Google analytics is report about 500 "current" users - so it is quite busy at that time.
>>>
>>>My question was about the affect of the Apache cache modules when
>>>running Lasso. Do you
>>have any experience with them?
>>>
>>>Regards
>>>jon
>>>
>>>-----Original Message-----
>>>From: [hidden email]
>>><[hidden email]> On Behalf Of Bil Corry
>>>Sent: 18 March 2019 03:49 PM
>>>To: [hidden email]
>>>Subject: Re: Lasso 8 on Apache
>>>
>>>You might want to see if the performance issue is on the Lasso side, or the Apache side.
>>>
>>>For Apache, you can log the total response time using %D (microseconds) or %T (seconds):
>>>
>>>http://httpd.apache.org/docs/current/mod/mod_log_config.html
>>>
>>>Within Lasso, log the processing time for each page using a timer (or
>>>just output it in a
>>comment in the HTML and load the page yourself).  My bet is the bulk
>>of the page load time is in Lasso, not Apache, which means optimizing
>>your pages in Lasso, or perhaps ensuring databases are not corrupted,
>>the disk isn't nearly full, the internal logs are not packed with
>>years of rows, etc.
>>>
>>>
>>>- Bil
>>>
>>>
>>>> On Mon, Mar 18, 2019 at 11:27 AM Jon Harris <[hidden email]> wrote:
>>>>   Hi
>>>>   We run Lasso 8 on a Windows VM under Apache 2.4.
>>>>    At peak times, (which is generally early evening) the site
>>>>really  slows down. We have been through our code adding "cache"
>>>>tags everywhere:
>>>>   Cache(-Name="headerCache", -Expires=120) ; // the code....
>>>> /cache ;
>>>> (etc, etc)
>>>>   - and the overall performance has improved.
>>>>    However, it is still lumpy (we are seeing a number of
>>>>408 errors in  the apache logs) We have tried adding additional
>>>>processors and RAM  to the VM - but this hasn't made any noticeable
>>>>difference. As Lasso  8 is only 32-bit (although the connector is
>>>>64-bit) there doesn't  seem any point allocating the server any more
>>>>RAM, than the 8Gb already allocated to it.
>>>>    Looking at httpd.conf, I noticed there are a number of cache and  
>>>>proxy modules, most of which are commented out on our (default)
>>>>installation.
>>>>    LoadModule cache_module modules/mod_cache.so LoadModule
>>>>cache_disk_module modules/mod_cache_disk.so LoadModule
>>>>cache_socache_module modules/mod_cache_socache.so LoadModule
>>>>proxy_fcgi_module modules/mod_proxy_fcgi.so LoadModule
>>>>proxy_hcheck_module modules/mod_proxy_hcheck.so LoadModule
>>>>proxy_html_module modules/mod_proxy_html.so LoadModule
>>>>proxy_http_module modules/mod_proxy_http.so LoadModule
>>>>proxy_scgi_module modules/mod_proxy_scgi.so LoadModule
>>>>proxy_uwsgi_module modules/mod_proxy_uwsgi.so LoadModule
>>>>proxy_wstunnel_module modules/mod_proxy_wstunnel.so
>>>>    Do any of these modules make an difference to the performance of
>>>>Lasso?
>>>>    Also, we have a dozen or so rewrite rules in our .htaccess.  Is
>>>>it  worth trying to ditch .htaccess and write those rules into the  
>>>>httpd.conf? My understanding is that for every file request Apache  
>>>>has to run the url through the .htaccess and as result it can run
>>>>faster with without
>>one.
>>>>    Beyond this, without major rewriting, would a front-end proxy
>>>>server  be the only viable solution.
>>>>   I'd be grateful for any useful advice.
>>>>   Regards
>>>> Jon Harris
>>>>    #############################################################
>>>>    This message is sent to you because you are subscribed to the  
>>>>mailing list Lasso [hidden email] Official list archives  
>>>>available at http://www.lassotalk.com To unsubscribe, E-mail
>>>> to: <[hidden email]>
>>>> Send administrative queries to  <[hidden email]>
>>>>
>>>
>>>#############################################################
>>>
>>>This message is sent to you because you are subscribed to the mailing
>>>list Lasso [hidden email] Official list archives available
>>>at http://www.lassotalk.com To unsubscribe, E-mail to:
>>><[hidden email]>
>>>Send administrative queries to  <[hidden email]>
>>>
>>>#############################################################
>>>
>>>This message is sent to you because you are subscribed to the mailing
>>>list Lasso [hidden email] Official list archives available
>>>at http://www.lassotalk.com To unsubscribe, E-mail to:
>>><[hidden email]>
>>>Send administrative queries to  <[hidden email]>
>>
>>
>>
>>#############################################################
>>
>>This message is sent to you because you are subscribed to the mailing
>>list Lasso [hidden email] Official list archives available
>>at http://www.lassotalk.com To unsubscribe, E-mail to:
>><[hidden email]>
>>Send administrative queries to  <[hidden email]>
>>
>>#############################################################
>>
>>This message is sent to you because you are subscribed to the mailing
>>list Lasso [hidden email] Official list archives available
>>at http://www.lassotalk.com To unsubscribe, E-mail to:
>><[hidden email]>
>>Send administrative queries to  <[hidden email]>
>
>-- -- -- -- -- -- -- -- -- -- -- -- -- -- -- -- -- -- -- -- -- --
>Steve Piercy              Website Builder              Eugene, OR
><[hidden email]>               <http://www.stevepiercy.com/>
>
>
>#############################################################
>
>This message is sent to you because you are subscribed to the mailing
>list Lasso [hidden email] Official list archives available
>at http://www.lassotalk.com To unsubscribe, E-mail to:
><[hidden email]>
>Send administrative queries to  <[hidden email]>
>
>#############################################################
>
>This message is sent to you because you are subscribed to the mailing
>list Lasso [hidden email] Official list archives available
>at http://www.lassotalk.com To unsubscribe, E-mail to:
><[hidden email]>
>Send administrative queries to  <[hidden email]>

-- -- -- -- -- -- -- -- -- -- -- -- -- -- -- -- -- -- -- -- -- --
Steve Piercy              Website Builder              Eugene, OR
<[hidden email]>               <http://www.stevepiercy.com/>


#############################################################

This message is sent to you because you are subscribed to
  the mailing list Lasso [hidden email] Official list archives available at http://www.lassotalk.com To unsubscribe, E-mail to: <[hidden email]>
Send administrative queries to  <[hidden email]>

#############################################################

This message is sent to you because you are subscribed to
  the mailing list Lasso [hidden email]
Official list archives available at http://www.lassotalk.com
To unsubscribe, E-mail to: <[hidden email]>
Send administrative queries to  <[hidden email]>
Reply | Threaded
Open this post in threaded view
|

Re: Lasso 8 on Apache

Ke Carlton-3
Hello Jon,

"isn't failing too badly it's just taking around a minute for the home
page" — this is failing badly. Such a request doesn't just affect the page
in question; it affects everything else also.

There are many strategies to address something like this; asides from
caching you can also asynchronously load in components of the page (in a
sequentially manner) that are particularly slow.

This even helps with the cache strategy as each of those requests /
components can be cached individually.

Create an empty test.lasso file and hit it with: ab -c50 -n500 — does it
crash?

Assuming not then it's all the stuff you've coded that's leading to crashes
(proportionate to load). I think there's still a Lasso 8.x version of
l-debug out there — that should help with identifying the slow portions of
code.







On Tue, 26 Mar 2019 at 02:00, Jon Harris <[hidden email]> wrote:

> Hi Steve
>
> Thanks for responding to my posting.Most of our CMS updates, where
> possible, write html which we [include] when we assemble the page - so
> probably not a query issue.
>
> Yes, I have used Lasso timings. The [cache] tag has made the biggest
> difference to how fast lasso is. As I said, I saved the generated html as a
> separate html file and stress tested that, all my testing told me that as a
> flat-HTML site - it would cope fine.
>
> Incidentally, as a test we moved to Google cloud-based hosted MySQL and it
> was around 2.5 times slower than our own cloud. (We have our own cloud
> servers, all our VMs talk to MySQL via a separate MPLS network. We have our
> own fibre-channel link to the EMC SAN.
>
> I am sure our code could be optimised, as it has grown organically over a
> number of years. If we were starting from scratch now, it would almost
> certainly be quite different.
>
> Every app will fail under load. Its' not possible to design computer
> systems to cope with an infinite amount of traffic.  If you have highly
> optimised code, using container objects. eventually, when you have filled a
> data centre with Kubernetes - your app will crash.
>
> In fact our solution isn't failing too badly it's just taking around a
> minute for the home page, plus some critical pages to load when the server
> gets very busy. I'm sure if Lasso 8 was recompiled as a 64bit process I am
> pretty sure it would copy much better, but I am not blaming the technology
> - it is what it is.
>
> I have decided to move to Nginx to proxy lasso requests, my early testing
> looks quite promising. I am just trying to get Nginx to serve the static
> content itself, as (this is not a Lasso issue) people are saying Nginx can
> serve static content much faster than Apache.
>
> Regards
> jon
>
> -----Original Message-----
> From: [hidden email] <[hidden email]>
> On Behalf Of Steve Piercy - Website Builder
> Sent: 21 March 2019 01:19 PM
> To: [hidden email]
> Subject: RE: Lasso 8 on Apache
>
> Have you enabled mysql slow query log and analyzed results?
> https://dev.mysql.com/doc/refman/8.0/en/slow-query-log.html
>
> Have you enabled a Lasso timer, logging timing, and analyzed results?
> http://www.lassosoft.com/tagswap/detail/timer
>
> Before you blame the architecture (where's your data to support your
> claim?) or pursue other remedies, I would look for the usual suspects: shit
> code and shit queries.  Everyone's code smells.  My code does not smell
> like roses, and I'm sure my legacy Lasso 8 apps would fall over under load.
>
> --steve
>
>
> On 3/21/19 at 9:39 AM, [hidden email] (Jon Harris) pronounced:
>
> >Hi Steve
> >
> >That is interesting idea - I hadn't thought of that as a solution. I
> >could zap the cache in the CMS and set a longer cache on the public
> >side.
> >
> >I have had a few problems with cache tags. For example, if we cache the
> >headers (which are fairly static) the page title is the one from the
> >first visitor to reload the cache. So I have created separate cache
> >headers on each page - which is less than optimal.
> >
> >The other problem is the page parameters. I have had to do things like:
> >
> >cache(-name="resultsCache" + action_param('param1')+
> >action_param('param2'),-expires=120) ;
> >
> >In the cms, I will see if I can loop these discreet params and delete
> >all the potential cache names.
> >
> >(Just to point out -  we don't use actually use "action_param"
> >- it's actually a sanitised VAR()  )
> >
> >I am looking at creating a "nginx.ourdomain.com" as a front-end proxy
> >server - to see how it will behave and how well we can control the
> >cache.
> >
> >Ultimately, I think the issue is down to the Lasso architecture.
> >Apache and the Lasso connector are 64-bit, but the lasso service is
> >32-bit, which is the likely performance limiter. In which case a
> >front-end solution feels like it will give us some future-proofing for
> >traffic spikes and growth.
> >
> >Regards
> >Jon Harris
> >
> >
> >
> >-----Original Message-----
> >From: [hidden email]
> ><[hidden email]> On Behalf Of Steve Piercy - Website
> >Builder
> >Sent: 19 March 2019 06:13 PM
> >To: [hidden email]
> >Subject: RE: Lasso 8 on Apache
> >
> >Hi Jon,
> >
> >When data is updated, you need to update the cache.  Why not do that
> >programmatically in Lasso instead of setting a low cache expiration
> >time?
> >
> >https://reference.lassosoft.com/LassoReference.LassoApp?[Cache_Delete]
> >https://reference.lassosoft.com/LassoReference.LassoApp?[Cache_Empty]
> >https://reference.lassosoft.com/LassoReference.LassoApp?[Cache_Store]
> >https://reference.lassosoft.com/LassoReference.LassoApp?[Cache_Fetch]
> >
> >You could add other caching mechanisms (web server, redis, CloudFlare,
> >etc.), but that adds complexity and layers of caching that make it
> >harder to debug.
> >
> >--steve
> >
> >
> >On 3/19/19 at 10:09 AM, [hidden email] (Jon Harris) pronounced:
> >
> >>Hi Mason
> >>
> >>Thanks for responding to my post.
> >>
> >>It's quite an interesting site -it is showing sports results.
> >>So we have 1000s of views of the same data in a relatively short
> >>period. Once the games are over, the data is fairly static. However,
> >>we want the maximum delay to be two minutes, so during games visitors
> >>can see a reasonably updated current score.  We have set the cache
> >>timeout to 120 seconds, meaning the average delay is one minute.
> >>As I type this, I am thinking that it might be possible to set the
> >>cache expiry dynamically higher - after a game has finished.
> >>How, do we  use Apache's caching instead of Lasso's. Do we remove the
> >>"cache" tags in our code and change something in the httpd.conf?
> >>
> >>I was thinking of creating an NGINX server as a proxy front-end
> >>server, or using proxy hosting service such as cloudFront, but this is
> >>quite a major project and we are concerned about how we can control
> >>the cache times.
> >>
> >>Regards
> >>Jon Harris
> >>
> >>-----Original Message-----
> >>From: [hidden email]
> >><[hidden email]> On Behalf Of Mason Miller
> >>Sent: 18 March 2019 09:44 PM
> >>To: [hidden email]
> >>Subject: Re: Lasso 8 on Apache
> >>
> >>How often does the data on the pages change?
> >>
> >>You might get a lot of mileage using Apache’s caching instead of
> >>Lasso’s. You might even set up a different file extension for the
> >>pages you want cached and those that need to be dynamic every time.
> >>Mason
> >>
> >>Sent from my iPhone
> >>
> >>>On Mar 18, 2019, at 12:55 PM, Jon Harris <[hidden email]> wrote:
> >>>
> >>>Hi Bill
> >>>
> >>>Thanks for the response.
> >>>
> >>>I saved the "generated" html as a straight .html page and ran a user
> >>>simulator on both
> >>the .html and .lasso files.
> >>>
> >>>It's not brilliant and I am sure there are better tools, but we are
> >>>using the "Paessler
> >>Webserver Stress Tool 8" to simulate users against 4 or 5 busy urls.
> >>From what we found the .html files can cope with the load and the
> >>.lasso versions of the same pages struggle.
> >>>
> >>>- So it does look like it is Lasso that's the bottleneck.
> >>>
> >>>We have taken mySQL out of the loop by using the [cache] tags around
> >>>the database calls,
> >>although for the most part we are using Lasso to write html - from the
> >>data in the CMS. Rather than use records loops directly on the page.
> >>>
> >>>Google analytics is report about 500 "current" users - so it is quite
> busy at that time.
> >>>
> >>>My question was about the affect of the Apache cache modules when
> >>>running Lasso. Do you
> >>have any experience with them?
> >>>
> >>>Regards
> >>>jon
> >>>
> >>>-----Original Message-----
> >>>From: [hidden email]
> >>><[hidden email]> On Behalf Of Bil Corry
> >>>Sent: 18 March 2019 03:49 PM
> >>>To: [hidden email]
> >>>Subject: Re: Lasso 8 on Apache
> >>>
> >>>You might want to see if the performance issue is on the Lasso side, or
> the Apache side.
> >>>
> >>>For Apache, you can log the total response time using %D (microseconds)
> or %T (seconds):
> >>>
> >>>http://httpd.apache.org/docs/current/mod/mod_log_config.html
> >>>
> >>>Within Lasso, log the processing time for each page using a timer (or
> >>>just output it in a
> >>comment in the HTML and load the page yourself).  My bet is the bulk
> >>of the page load time is in Lasso, not Apache, which means optimizing
> >>your pages in Lasso, or perhaps ensuring databases are not corrupted,
> >>the disk isn't nearly full, the internal logs are not packed with
> >>years of rows, etc.
> >>>
> >>>
> >>>- Bil
> >>>
> >>>
> >>>> On Mon, Mar 18, 2019 at 11:27 AM Jon Harris <[hidden email]>
> wrote:
> >>>>   Hi
> >>>>   We run Lasso 8 on a Windows VM under Apache 2.4.
> >>>>    At peak times, (which is generally early evening) the site
> >>>>really  slows down. We have been through our code adding "cache"
> >>>>tags everywhere:
> >>>>   Cache(-Name="headerCache", -Expires=120) ; // the code....
> >>>> /cache ;
> >>>> (etc, etc)
> >>>>   - and the overall performance has improved.
> >>>>    However, it is still lumpy (we are seeing a number of
> >>>>408 errors in  the apache logs) We have tried adding additional
> >>>>processors and RAM  to the VM - but this hasn't made any noticeable
> >>>>difference. As Lasso  8 is only 32-bit (although the connector is
> >>>>64-bit) there doesn't  seem any point allocating the server any more
> >>>>RAM, than the 8Gb already allocated to it.
> >>>>    Looking at httpd.conf, I noticed there are a number of cache and
> >>>>proxy modules, most of which are commented out on our (default)
> >>>>installation.
> >>>>    LoadModule cache_module modules/mod_cache.so LoadModule
> >>>>cache_disk_module modules/mod_cache_disk.so LoadModule
> >>>>cache_socache_module modules/mod_cache_socache.so LoadModule
> >>>>proxy_fcgi_module modules/mod_proxy_fcgi.so LoadModule
> >>>>proxy_hcheck_module modules/mod_proxy_hcheck.so LoadModule
> >>>>proxy_html_module modules/mod_proxy_html.so LoadModule
> >>>>proxy_http_module modules/mod_proxy_http.so LoadModule
> >>>>proxy_scgi_module modules/mod_proxy_scgi.so LoadModule
> >>>>proxy_uwsgi_module modules/mod_proxy_uwsgi.so LoadModule
> >>>>proxy_wstunnel_module modules/mod_proxy_wstunnel.so
> >>>>    Do any of these modules make an difference to the performance of
> >>>>Lasso?
> >>>>    Also, we have a dozen or so rewrite rules in our .htaccess.  Is
> >>>>it  worth trying to ditch .htaccess and write those rules into the
> >>>>httpd.conf? My understanding is that for every file request Apache
> >>>>has to run the url through the .htaccess and as result it can run
> >>>>faster with without
> >>one.
> >>>>    Beyond this, without major rewriting, would a front-end proxy
> >>>>server  be the only viable solution.
> >>>>   I'd be grateful for any useful advice.
> >>>>   Regards
> >>>> Jon Harris
> >>>>    #############################################################
> >>>>    This message is sent to you because you are subscribed to the
> >>>>mailing list Lasso [hidden email] Official list archives
> >>>>available at http://www.lassotalk.com To unsubscribe, E-mail
> >>>> to: <[hidden email]>
> >>>> Send administrative queries to  <[hidden email]>
> >>>>
> >>>
> >>>#############################################################
> >>>
> >>>This message is sent to you because you are subscribed to the mailing
> >>>list Lasso [hidden email] Official list archives available
> >>>at http://www.lassotalk.com To unsubscribe, E-mail to:
> >>><[hidden email]>
> >>>Send administrative queries to  <[hidden email]>
> >>>
> >>>#############################################################
> >>>
> >>>This message is sent to you because you are subscribed to the mailing
> >>>list Lasso [hidden email] Official list archives available
> >>>at http://www.lassotalk.com To unsubscribe, E-mail to:
> >>><[hidden email]>
> >>>Send administrative queries to  <[hidden email]>
> >>
> >>
> >>
> >>#############################################################
> >>
> >>This message is sent to you because you are subscribed to the mailing
> >>list Lasso [hidden email] Official list archives available
> >>at http://www.lassotalk.com To unsubscribe, E-mail to:
> >><[hidden email]>
> >>Send administrative queries to  <[hidden email]>
> >>
> >>#############################################################
> >>
> >>This message is sent to you because you are subscribed to the mailing
> >>list Lasso [hidden email] Official list archives available
> >>at http://www.lassotalk.com To unsubscribe, E-mail to:
> >><[hidden email]>
> >>Send administrative queries to  <[hidden email]>
> >
> >-- -- -- -- -- -- -- -- -- -- -- -- -- -- -- -- -- -- -- -- -- --
> >Steve Piercy              Website Builder              Eugene, OR
> ><[hidden email]>               <http://www.stevepiercy.com/>
> >
> >
> >#############################################################
> >
> >This message is sent to you because you are subscribed to the mailing
> >list Lasso [hidden email] Official list archives available
> >at http://www.lassotalk.com To unsubscribe, E-mail to:
> ><[hidden email]>
> >Send administrative queries to  <[hidden email]>
> >
> >#############################################################
> >
> >This message is sent to you because you are subscribed to the mailing
> >list Lasso [hidden email] Official list archives available
> >at http://www.lassotalk.com To unsubscribe, E-mail to:
> ><[hidden email]>
> >Send administrative queries to  <[hidden email]>
>
> -- -- -- -- -- -- -- -- -- -- -- -- -- -- -- -- -- -- -- -- -- --
> Steve Piercy              Website Builder              Eugene, OR
> <[hidden email]>               <http://www.stevepiercy.com/>
>
>
> #############################################################
>
> This message is sent to you because you are subscribed to
>   the mailing list Lasso [hidden email] Official list archives
> available at http://www.lassotalk.com To unsubscribe, E-mail to: <
> [hidden email]>
> Send administrative queries to  <[hidden email]>
>
> #############################################################
>
> This message is sent to you because you are subscribed to
>   the mailing list Lasso [hidden email]
> Official list archives available at http://www.lassotalk.com
> To unsubscribe, E-mail to: <[hidden email]>
> Send administrative queries to  <[hidden email]>

#############################################################

This message is sent to you because you are subscribed to
  the mailing list Lasso [hidden email]
Official list archives available at http://www.lassotalk.com
To unsubscribe, E-mail to: <[hidden email]>
Send administrative queries to  <[hidden email]>
Reply | Threaded
Open this post in threaded view
|

RE: Lasso 8 on Apache

Jon Harris-2
Hi List

Thanks to all the replies to my query.

I have now got Nginx working as a proxy-pass server. I have also created an smb share so Nginx is serving static content directly, without having to bother Apache.

We are using an MPLS connection to the Apache server (they are separate VMs), which means we don't have to use ssl, which should also mean less load on Apache.

In nginx we added the real IP to the headers, then back on Apache we enabled "mod_remoteip.so" and added "{X-Forwarded-For}" to our logging. This means we see the real-ip address of the visitor in the Apache log files.

Initial tests look good, so I am hopeful this is a solution.

Jon Harris




-----Original Message-----
From: [hidden email] <[hidden email]> On Behalf Of Ke Carlton
Sent: 25 March 2019 10:55 PM
To: Lasso Talk <[hidden email]>
Subject: Re: Lasso 8 on Apache

Hello Jon,

"isn't failing too badly it's just taking around a minute for the home page" — this is failing badly. Such a request doesn't just affect the page in question; it affects everything else also.

There are many strategies to address something like this; asides from caching you can also asynchronously load in components of the page (in a sequentially manner) that are particularly slow.

This even helps with the cache strategy as each of those requests / components can be cached individually.

Create an empty test.lasso file and hit it with: ab -c50 -n500 — does it crash?

Assuming not then it's all the stuff you've coded that's leading to crashes (proportionate to load). I think there's still a Lasso 8.x version of l-debug out there — that should help with identifying the slow portions of code.







On Tue, 26 Mar 2019 at 02:00, Jon Harris <[hidden email]> wrote:

> Hi Steve
>
> Thanks for responding to my posting.Most of our CMS updates, where
> possible, write html which we [include] when we assemble the page - so
> probably not a query issue.
>
> Yes, I have used Lasso timings. The [cache] tag has made the biggest
> difference to how fast lasso is. As I said, I saved the generated html
> as a separate html file and stress tested that, all my testing told me
> that as a flat-HTML site - it would cope fine.
>
> Incidentally, as a test we moved to Google cloud-based hosted MySQL
> and it was around 2.5 times slower than our own cloud. (We have our
> own cloud servers, all our VMs talk to MySQL via a separate MPLS
> network. We have our own fibre-channel link to the EMC SAN.
>
> I am sure our code could be optimised, as it has grown organically
> over a number of years. If we were starting from scratch now, it would
> almost certainly be quite different.
>
> Every app will fail under load. Its' not possible to design computer
> systems to cope with an infinite amount of traffic.  If you have
> highly optimised code, using container objects. eventually, when you
> have filled a data centre with Kubernetes - your app will crash.
>
> In fact our solution isn't failing too badly it's just taking around a
> minute for the home page, plus some critical pages to load when the
> server gets very busy. I'm sure if Lasso 8 was recompiled as a 64bit
> process I am pretty sure it would copy much better, but I am not
> blaming the technology
> - it is what it is.
>
> I have decided to move to Nginx to proxy lasso requests, my early
> testing looks quite promising. I am just trying to get Nginx to serve
> the static content itself, as (this is not a Lasso issue) people are
> saying Nginx can serve static content much faster than Apache.
>
> Regards
> jon
>
> -----Original Message-----
> From: [hidden email]
> <[hidden email]>
> On Behalf Of Steve Piercy - Website Builder
> Sent: 21 March 2019 01:19 PM
> To: [hidden email]
> Subject: RE: Lasso 8 on Apache
>
> Have you enabled mysql slow query log and analyzed results?
> https://dev.mysql.com/doc/refman/8.0/en/slow-query-log.html
>
> Have you enabled a Lasso timer, logging timing, and analyzed results?
> http://www.lassosoft.com/tagswap/detail/timer
>
> Before you blame the architecture (where's your data to support your
> claim?) or pursue other remedies, I would look for the usual suspects:
> shit code and shit queries.  Everyone's code smells.  My code does not
> smell like roses, and I'm sure my legacy Lasso 8 apps would fall over under load.
>
> --steve
>
>
> On 3/21/19 at 9:39 AM, [hidden email] (Jon Harris) pronounced:
>
> >Hi Steve
> >
> >That is interesting idea - I hadn't thought of that as a solution. I
> >could zap the cache in the CMS and set a longer cache on the public
> >side.
> >
> >I have had a few problems with cache tags. For example, if we cache
> >the headers (which are fairly static) the page title is the one from
> >the first visitor to reload the cache. So I have created separate
> >cache headers on each page - which is less than optimal.
> >
> >The other problem is the page parameters. I have had to do things like:
> >
> >cache(-name="resultsCache" + action_param('param1')+
> >action_param('param2'),-expires=120) ;
> >
> >In the cms, I will see if I can loop these discreet params and delete
> >all the potential cache names.
> >
> >(Just to point out -  we don't use actually use "action_param"
> >- it's actually a sanitised VAR()  )
> >
> >I am looking at creating a "nginx.ourdomain.com" as a front-end proxy
> >server - to see how it will behave and how well we can control the
> >cache.
> >
> >Ultimately, I think the issue is down to the Lasso architecture.
> >Apache and the Lasso connector are 64-bit, but the lasso service is
> >32-bit, which is the likely performance limiter. In which case a
> >front-end solution feels like it will give us some future-proofing
> >for traffic spikes and growth.
> >
> >Regards
> >Jon Harris
> >
> >
> >
> >-----Original Message-----
> >From: [hidden email]
> ><[hidden email]> On Behalf Of Steve Piercy -
> >Website Builder
> >Sent: 19 March 2019 06:13 PM
> >To: [hidden email]
> >Subject: RE: Lasso 8 on Apache
> >
> >Hi Jon,
> >
> >When data is updated, you need to update the cache.  Why not do that
> >programmatically in Lasso instead of setting a low cache expiration
> >time?
> >
> >https://reference.lassosoft.com/LassoReference.LassoApp?[Cache_Delete
> >]
> >https://reference.lassosoft.com/LassoReference.LassoApp?[Cache_Empty]
> >https://reference.lassosoft.com/LassoReference.LassoApp?[Cache_Store]
> >https://reference.lassosoft.com/LassoReference.LassoApp?[Cache_Fetch]
> >
> >You could add other caching mechanisms (web server, redis,
> >CloudFlare, etc.), but that adds complexity and layers of caching
> >that make it harder to debug.
> >
> >--steve
> >
> >
> >On 3/19/19 at 10:09 AM, [hidden email] (Jon Harris) pronounced:
> >
> >>Hi Mason
> >>
> >>Thanks for responding to my post.
> >>
> >>It's quite an interesting site -it is showing sports results.
> >>So we have 1000s of views of the same data in a relatively short
> >>period. Once the games are over, the data is fairly static. However,
> >>we want the maximum delay to be two minutes, so during games
> >>visitors can see a reasonably updated current score.  We have set
> >>the cache timeout to 120 seconds, meaning the average delay is one minute.
> >>As I type this, I am thinking that it might be possible to set the
> >>cache expiry dynamically higher - after a game has finished.
> >>How, do we  use Apache's caching instead of Lasso's. Do we remove
> >>the "cache" tags in our code and change something in the httpd.conf?
> >>
> >>I was thinking of creating an NGINX server as a proxy front-end
> >>server, or using proxy hosting service such as cloudFront, but this
> >>is quite a major project and we are concerned about how we can
> >>control the cache times.
> >>
> >>Regards
> >>Jon Harris
> >>
> >>-----Original Message-----
> >>From: [hidden email]
> >><[hidden email]> On Behalf Of Mason Miller
> >>Sent: 18 March 2019 09:44 PM
> >>To: [hidden email]
> >>Subject: Re: Lasso 8 on Apache
> >>
> >>How often does the data on the pages change?
> >>
> >>You might get a lot of mileage using Apache’s caching instead of
> >>Lasso’s. You might even set up a different file extension for the
> >>pages you want cached and those that need to be dynamic every time.
> >>Mason
> >>
> >>Sent from my iPhone
> >>
> >>>On Mar 18, 2019, at 12:55 PM, Jon Harris <[hidden email]> wrote:
> >>>
> >>>Hi Bill
> >>>
> >>>Thanks for the response.
> >>>
> >>>I saved the "generated" html as a straight .html page and ran a
> >>>user simulator on both
> >>the .html and .lasso files.
> >>>
> >>>It's not brilliant and I am sure there are better tools, but we are
> >>>using the "Paessler
> >>Webserver Stress Tool 8" to simulate users against 4 or 5 busy urls.
> >>From what we found the .html files can cope with the load and the
> >>.lasso versions of the same pages struggle.
> >>>
> >>>- So it does look like it is Lasso that's the bottleneck.
> >>>
> >>>We have taken mySQL out of the loop by using the [cache] tags
> >>>around the database calls,
> >>although for the most part we are using Lasso to write html - from
> >>the data in the CMS. Rather than use records loops directly on the page.
> >>>
> >>>Google analytics is report about 500 "current" users - so it is
> >>>quite
> busy at that time.
> >>>
> >>>My question was about the affect of the Apache cache modules when
> >>>running Lasso. Do you
> >>have any experience with them?
> >>>
> >>>Regards
> >>>jon
> >>>
> >>>-----Original Message-----
> >>>From: [hidden email]
> >>><[hidden email]> On Behalf Of Bil Corry
> >>>Sent: 18 March 2019 03:49 PM
> >>>To: [hidden email]
> >>>Subject: Re: Lasso 8 on Apache
> >>>
> >>>You might want to see if the performance issue is on the Lasso
> >>>side, or
> the Apache side.
> >>>
> >>>For Apache, you can log the total response time using %D
> >>>(microseconds)
> or %T (seconds):
> >>>
> >>>http://httpd.apache.org/docs/current/mod/mod_log_config.html
> >>>
> >>>Within Lasso, log the processing time for each page using a timer
> >>>(or just output it in a
> >>comment in the HTML and load the page yourself).  My bet is the bulk
> >>of the page load time is in Lasso, not Apache, which means
> >>optimizing your pages in Lasso, or perhaps ensuring databases are
> >>not corrupted, the disk isn't nearly full, the internal logs are not
> >>packed with years of rows, etc.
> >>>
> >>>
> >>>- Bil
> >>>
> >>>
> >>>> On Mon, Mar 18, 2019 at 11:27 AM Jon Harris
> >>>> <[hidden email]>
> wrote:
> >>>>   Hi
> >>>>   We run Lasso 8 on a Windows VM under Apache 2.4.
> >>>>    At peak times, (which is generally early evening) the site
> >>>>really  slows down. We have been through our code adding "cache"
> >>>>tags everywhere:
> >>>>   Cache(-Name="headerCache", -Expires=120) ; // the code....
> >>>> /cache ;
> >>>> (etc, etc)
> >>>>   - and the overall performance has improved.
> >>>>    However, it is still lumpy (we are seeing a number of
> >>>>408 errors in  the apache logs) We have tried adding additional
> >>>>processors and RAM  to the VM - but this hasn't made any
> >>>>noticeable difference. As Lasso  8 is only 32-bit (although the
> >>>>connector is
> >>>>64-bit) there doesn't  seem any point allocating the server any
> >>>>more RAM, than the 8Gb already allocated to it.
> >>>>    Looking at httpd.conf, I noticed there are a number of cache
> >>>>and proxy modules, most of which are commented out on our
> >>>>(default) installation.
> >>>>    LoadModule cache_module modules/mod_cache.so LoadModule
> >>>>cache_disk_module modules/mod_cache_disk.so LoadModule
> >>>>cache_socache_module modules/mod_cache_socache.so LoadModule
> >>>>proxy_fcgi_module modules/mod_proxy_fcgi.so LoadModule
> >>>>proxy_hcheck_module modules/mod_proxy_hcheck.so LoadModule
> >>>>proxy_html_module modules/mod_proxy_html.so LoadModule
> >>>>proxy_http_module modules/mod_proxy_http.so LoadModule
> >>>>proxy_scgi_module modules/mod_proxy_scgi.so LoadModule
> >>>>proxy_uwsgi_module modules/mod_proxy_uwsgi.so LoadModule
> >>>>proxy_wstunnel_module modules/mod_proxy_wstunnel.so
> >>>>    Do any of these modules make an difference to the performance
> >>>>of Lasso?
> >>>>    Also, we have a dozen or so rewrite rules in our .htaccess.  
> >>>>Is it  worth trying to ditch .htaccess and write those rules into
> >>>>the httpd.conf? My understanding is that for every file request
> >>>>Apache has to run the url through the .htaccess and as result it
> >>>>can run faster with without
> >>one.
> >>>>    Beyond this, without major rewriting, would a front-end proxy
> >>>>server  be the only viable solution.
> >>>>   I'd be grateful for any useful advice.
> >>>>   Regards
> >>>> Jon Harris
> >>>>    #############################################################
> >>>>    This message is sent to you because you are subscribed to the
> >>>>mailing list Lasso [hidden email] Official list
> >>>>archives available at http://www.lassotalk.com To unsubscribe,
> >>>>E-mail
> >>>> to: <[hidden email]>
> >>>> Send administrative queries to  
> >>>><[hidden email]>
> >>>>
> >>>
> >>>#############################################################
> >>>
> >>>This message is sent to you because you are subscribed to the
> >>>mailing list Lasso [hidden email] Official list archives
> >>>available at http://www.lassotalk.com To unsubscribe, E-mail to:
> >>><[hidden email]>
> >>>Send administrative queries to  <[hidden email]>
> >>>
> >>>#############################################################
> >>>
> >>>This message is sent to you because you are subscribed to the
> >>>mailing list Lasso [hidden email] Official list archives
> >>>available at http://www.lassotalk.com To unsubscribe, E-mail to:
> >>><[hidden email]>
> >>>Send administrative queries to  <[hidden email]>
> >>
> >>
> >>
> >>#############################################################
> >>
> >>This message is sent to you because you are subscribed to the
> >>mailing list Lasso [hidden email] Official list archives
> >>available at http://www.lassotalk.com To unsubscribe, E-mail to:
> >><[hidden email]>
> >>Send administrative queries to  <[hidden email]>
> >>
> >>#############################################################
> >>
> >>This message is sent to you because you are subscribed to the
> >>mailing list Lasso [hidden email] Official list archives
> >>available at http://www.lassotalk.com To unsubscribe, E-mail to:
> >><[hidden email]>
> >>Send administrative queries to  <[hidden email]>
> >
> >-- -- -- -- -- -- -- -- -- -- -- -- -- -- -- -- -- -- -- -- -- --
> >Steve Piercy              Website Builder              Eugene, OR
> ><[hidden email]>               <http://www.stevepiercy.com/>
> >
> >
> >#############################################################
> >
> >This message is sent to you because you are subscribed to the mailing
> >list Lasso [hidden email] Official list archives available
> >at http://www.lassotalk.com To unsubscribe, E-mail to:
> ><[hidden email]>
> >Send administrative queries to  <[hidden email]>
> >
> >#############################################################
> >
> >This message is sent to you because you are subscribed to the mailing
> >list Lasso [hidden email] Official list archives available
> >at http://www.lassotalk.com To unsubscribe, E-mail to:
> ><[hidden email]>
> >Send administrative queries to  <[hidden email]>
>
> -- -- -- -- -- -- -- -- -- -- -- -- -- -- -- -- -- -- -- -- -- --
> Steve Piercy              Website Builder              Eugene, OR
> <[hidden email]>               <http://www.stevepiercy.com/>
>
>
> #############################################################
>
> This message is sent to you because you are subscribed to
>   the mailing list Lasso [hidden email] Official list
> archives available at http://www.lassotalk.com To unsubscribe, E-mail
> to: < [hidden email]>
> Send administrative queries to  <[hidden email]>
>
> #############################################################
>
> This message is sent to you because you are subscribed to
>   the mailing list Lasso [hidden email] Official list
> archives available at http://www.lassotalk.com To unsubscribe, E-mail
> to: <[hidden email]>
> Send administrative queries to  <[hidden email]>

#############################################################

This message is sent to you because you are subscribed to
  the mailing list Lasso [hidden email] Official list archives available at http://www.lassotalk.com To unsubscribe, E-mail to: <[hidden email]>
Send administrative queries to  <[hidden email]>

#############################################################

This message is sent to you because you are subscribed to
  the mailing list Lasso [hidden email]
Official list archives available at http://www.lassotalk.com
To unsubscribe, E-mail to: <[hidden email]>
Send administrative queries to  <[hidden email]>