[OFBiz] Users - Problems with a site that was functioning fine ...

[Ian Gilbert]

Hi All,

I have a situation where one of our sites (http://www.silverchilli.com)
isn't responding in a timely fashion.  This is a build from about fourteen
months ago which has been fairly stable for a good long time.  However
yesterday it appeared in a newspaper article and, of course, since then
I've been noticing problems with the performance, responsiveness and
errors being generated in strange places.

I've been ploughing through the logs and have noticed a *huge* number of
error messages.  The generation of these would certainly have adverse
impact on performance and may result in some of the timeouts that we've
been having.

My problem is that, although I set up and configured the original build,
it was then subject to a series of mods over which I had no control.
These are, in the most part, UI mods (commenting out chunks of
functionality) and so shouldn't have fed thru to the remainder of the app
but there may have been some logic mods (although this seems unlikely).
These mods haven't been done recently (at least a year old in the most
recent case) and so are not directly responsible for the current
instability.

If possible I'd like to avoid spending a huge amount of time on this atm
bearing in mind that I'm probably going to upgrade the whole system once
we've got a stable release to go to and I've sorted our principle system.
Since most of the error messages (on the client side) seem to be related
to processes timing out I was wondering if anyone new of any ways to
improve the performance (I think that in previous incarnations I disabled
error logging for example) that might make a difference in this case.  I
am at a bit of a loss to understand the problem here as it has been
running well for some time.

I have restarted both the Ofbiz service and Postgres (version 7.3 which
wouldn't shut down via a command and required that I forcibly reboot the
box which I was dissapointed by but is doesn't relate to Ofbiz) and indeed
the server that these are running on.

I haven't been sure wether to attach the error logs as I think that they
are made much more complex by the existence of these mods which I can't
easily explain.  There are a large number of warnings (which I ignore) but
the errors seem to refer to a missing Minerva source (which surely would
have been a problem before) and then lots of attempts to rollback
transactions.  It was the existence of these transactions and the seeming
inability of Postgres to sort them out which caused me to reboot the
server.  For a time it looked like it had worked but it has deteriorated.
I decided to link to  the logs in the end although these are fairly
substantial.  If you are would like to view them then they can be aquired
at http://www.ethicalshopper.co.uk/~ian/logs.zip

Does anyone have any ideas how I might improve the performance of this?  I
believe that there are so many errors being generated that it's timing out
the webapp.  Removing the error logging (if this is possible) is certainly
an undesirable hack but the guys are, understandably, keen to make the
most of the newspaper exposure but the poor performance/inability to
login/inability to order has resulted in my having to place a temporary
error message on their site.  It's a shame as this is the first problem
they've had for 204 days (and then only for a couple of hours) and it's
typical that it's co-incided with the newspaper review.

The box is reasonably spec'd (an IBM box with, I think, a 700 mhz
processor and 256MB ram.  The postgres database isn't huge (~70MB which
compares to a 200MB ethicalshopper db which is performing fine).  The disk
isn't full (less than 20% used) although the box does seem sluggish when
logged in.

Thanks for any help anyone can offer.

Very best wishes

Ian

 
_______________________________________________
Users mailing list
[hidden email]
http://lists.ofbiz.org/mailman/listinfo/users

[Brett]

Ian,

This is hard to trouble shoot from your explanation.  I would suggest
posting some of the common error stack traces you are seeing.

OFBiz like most enterprise applications takes a lot of support to keep
it running.  You can't expect the application to behave the same over
time because a lot is changing, particularly on the database.

You might want to look at some of your tables to see how big they are
getting.  We have run into problems with the job_sandbox table growing
too big because some of the automated archiving services weren't
running.

Hope that helps,


Brett

On 8/16/05, Ian Gilbert <[hidden email]> wrote:

> Hi All,
>
> I have a situation where one of our sites (http://www.silverchilli.com)
> isn't responding in a timely fashion.  This is a build from about fourteen
> months ago which has been fairly stable for a good long time.  However
> yesterday it appeared in a newspaper article and, of course, since then
> I've been noticing problems with the performance, responsiveness and
> errors being generated in strange places.
>
> I've been ploughing through the logs and have noticed a *huge* number of
> error messages.  The generation of these would certainly have adverse
> impact on performance and may result in some of the timeouts that we've
> been having.
>
> My problem is that, although I set up and configured the original build,
> it was then subject to a series of mods over which I had no control.
> These are, in the most part, UI mods (commenting out chunks of
> functionality) and so shouldn't have fed thru to the remainder of the app
> but there may have been some logic mods (although this seems unlikely).
> These mods haven't been done recently (at least a year old in the most
> recent case) and so are not directly responsible for the current
> instability.
>
> If possible I'd like to avoid spending a huge amount of time on this atm
> bearing in mind that I'm probably going to upgrade the whole system once
> we've got a stable release to go to and I've sorted our principle system.
> Since most of the error messages (on the client side) seem to be related
> to processes timing out I was wondering if anyone new of any ways to
> improve the performance (I think that in previous incarnations I disabled
> error logging for example) that might make a difference in this case.  I
> am at a bit of a loss to understand the problem here as it has been
> running well for some time.
>
> I have restarted both the Ofbiz service and Postgres (version 7.3 which
> wouldn't shut down via a command and required that I forcibly reboot the
> box which I was dissapointed by but is doesn't relate to Ofbiz) and indeed
> the server that these are running on.
>
> I haven't been sure wether to attach the error logs as I think that they
> are made much more complex by the existence of these mods which I can't
> easily explain.  There are a large number of warnings (which I ignore) but
> the errors seem to refer to a missing Minerva source (which surely would
> have been a problem before) and then lots of attempts to rollback
> transactions.  It was the existence of these transactions and the seeming
> inability of Postgres to sort them out which caused me to reboot the
> server.  For a time it looked like it had worked but it has deteriorated.
> I decided to link to  the logs in the end although these are fairly
> substantial.  If you are would like to view them then they can be aquired
> at http://www.ethicalshopper.co.uk/~ian/logs.zip
>
> Does anyone have any ideas how I might improve the performance of this?  I
> believe that there are so many errors being generated that it's timing out
> the webapp.  Removing the error logging (if this is possible) is certainly
> an undesirable hack but the guys are, understandably, keen to make the
> most of the newspaper exposure but the poor performance/inability to
> login/inability to order has resulted in my having to place a temporary
> error message on their site.  It's a shame as this is the first problem
> they've had for 204 days (and then only for a couple of hours) and it's
> typical that it's co-incided with the newspaper review.
>
> The box is reasonably spec'd (an IBM box with, I think, a 700 mhz
> processor and 256MB ram.  The postgres database isn't huge (~70MB which
> compares to a 200MB ethicalshopper db which is performing fine).  The disk
> isn't full (less than 20% used) although the box does seem sluggish when
> logged in.
>
> Thanks for any help anyone can offer.
>
> Very best wishes
>
> Ian
>
>
> _______________________________________________
> Users mailing list
> [hidden email]
> http://lists.ofbiz.org/mailman/listinfo/users
>

 
_______________________________________________
Users mailing list
[hidden email]
http://lists.ofbiz.org/mailman/listinfo/users

[Ian Gilbert]

Thanks for that Brett.  I've taken a look through the various traces now
and I think that I'm making some progress.  I am trying out a few things
(removing the Shark app) and am going to see how this works.  There are a
load of errors but I'm not sure how many of these at the moment are
related to the system and how many are a result of the users mods.  In
addition I'd like to see which ones look critical and which ones are just
'warnings.  I will wait until I have a clearer idea of the problem before
posting these up and try to refine them down a bit otherwise we will be in
a 'can't see the wood for the trees' situation.

Thanks again for the pointers.  That's been helpful.  I hope to sort this
but otherwise I'll be back in the morning.  It's late here and I'm
thinking bed is now a good option :)

Thanks again.

Speak to you soon

Ian

On Tue, August 16, 2005 3:32 pm, Brett Palmer wrote:

> Ian,
>
>
> This is hard to trouble shoot from your explanation.  I would suggest
> posting some of the common error stack traces you are seeing.
>
> OFBiz like most enterprise applications takes a lot of support to keep
> it running.  You can't expect the application to behave the same over time
> because a lot is changing, particularly on the database.
>
> You might want to look at some of your tables to see how big they are
> getting.  We have run into problems with the job_sandbox table growing too
> big because some of the automated archiving services weren't running.
>
> Hope that helps,
>
>
>
> Brett
>
>
> On 8/16/05, Ian Gilbert <[hidden email]> wrote:
>
>> Hi All,
>>
>>
>> I have a situation where one of our sites (http://www.silverchilli.com)
>>  isn't responding in a timely fashion.  This is a build from about
>> fourteen months ago which has been fairly stable for a good long time.
>> However
>> yesterday it appeared in a newspaper article and, of course, since then
>> I've been noticing problems with the performance, responsiveness and
>> errors being generated in strange places.
>>
>> I've been ploughing through the logs and have noticed a *huge* number
>> of error messages.  The generation of these would certainly have adverse
>>  impact on performance and may result in some of the timeouts that
>> we've been having.
>>
>> My problem is that, although I set up and configured the original
>> build, it was then subject to a series of mods over which I had no
>> control. These are, in the most part, UI mods (commenting out chunks of
>> functionality) and so shouldn't have fed thru to the remainder of the
>> app but there may have been some logic mods (although this seems
>> unlikely). These mods haven't been done recently (at least a year old in
>> the most recent case) and so are not directly responsible for the
>> current instability.
>>
>> If possible I'd like to avoid spending a huge amount of time on this
>> atm bearing in mind that I'm probably going to upgrade the whole system
>> once we've got a stable release to go to and I've sorted our principle
>> system. Since most of the error messages (on the client side) seem to be
>> related to processes timing out I was wondering if anyone new of any
>> ways to improve the performance (I think that in previous incarnations I
>> disabled error logging for example) that might make a difference in this
>> case.  I am at a bit of a loss to understand the problem here as it has
>> been running well for some time.
>>
>> I have restarted both the Ofbiz service and Postgres (version 7.3 which
>>  wouldn't shut down via a command and required that I forcibly reboot
>> the box which I was dissapointed by but is doesn't relate to Ofbiz) and
>> indeed the server that these are running on.
>>
>> I haven't been sure wether to attach the error logs as I think that
>> they are made much more complex by the existence of these mods which I
>> can't easily explain.  There are a large number of warnings (which I
>> ignore) but the errors seem to refer to a missing Minerva source (which
>> surely would have been a problem before) and then lots of attempts to
>> rollback transactions.  It was the existence of these transactions and
>> the seeming inability of Postgres to sort them out which caused me to
>> reboot the server.  For a time it looked like it had worked but it has
>> deteriorated. I decided to link to  the logs in the end although these
>> are fairly substantial.  If you are would like to view them then they
>> can be aquired at http://www.ethicalshopper.co.uk/~ian/logs.zip
>>
>> Does anyone have any ideas how I might improve the performance of this?
>> I
>> believe that there are so many errors being generated that it's timing
>> out the webapp.  Removing the error logging (if this is possible) is
>> certainly an undesirable hack but the guys are, understandably, keen to
>> make the most of the newspaper exposure but the poor
>> performance/inability to login/inability to order has resulted in my
>> having to place a temporary error message on their site.  It's a shame
>> as this is the first problem they've had for 204 days (and then only for
>> a couple of hours) and it's typical that it's co-incided with the
>> newspaper review.
>>
>> The box is reasonably spec'd (an IBM box with, I think, a 700 mhz
>> processor and 256MB ram.  The postgres database isn't huge (~70MB which
>> compares to a 200MB ethicalshopper db which is performing fine).  The
>> disk isn't full (less than 20% used) although the box does seem sluggish
>> when logged in.
>>
>> Thanks for any help anyone can offer.
>>
>>
>> Very best wishes
>>
>>
>> Ian
>>
>>
>>
>> _______________________________________________
>> Users mailing list
>> [hidden email]
>> http://lists.ofbiz.org/mailman/listinfo/users
>>
>>
>
> _______________________________________________
> Users mailing list
> [hidden email]
> http://lists.ofbiz.org/mailman/listinfo/users
>
>

 
_______________________________________________
Users mailing list
[hidden email]
http://lists.ofbiz.org/mailman/listinfo/users