How to debug "An internal server error occurred. Please try again later."?

[Adam Ryczkowski]

On one of my seemingly identical virtual severs I always receive an
error message "An internal server error occurred. Please try again
later." after commiting change or posting a new bug with http user
interface. The data commits just fine, but the error message is annoying
and definitely doesn't look good. And commiting with Mylyn (Eclipse) is
broken, since Mylyn treats this error message as error and assumes the
actual data in fact is not commited.

There is nothing on apache2/error.log and mysql.err.

I hope there is a way to make Bugzilla give more meaningfull error
reports. I just couldn't find how to enable it.

* * *

I use the latest Bugzilla version downloaded with "bzr co
bzr://bzr.mozilla.org/bugzilla/trunk bugzilla" and subsequent upgrades
with "bzr update" and successfull "./checksetup.pl".

Bugzilla 4.5 on standard LXC container on Ubuntu Quantal server with
kernel "3.7.6-030706-generic".

P.S. This is a repost. I believe the previous mail sent on 17.II.2013
went to mozilla.org's spam bin.
--

Adam Ryczkowski
+48505919892 <callto:+48505919892>
Skype:sisteczko <skype:sisteczko>

[Thorsten Schöning]

Guten Tag Adam Ryczkowski,
am Montag, 18. Februar 2013 um 20:17 schrieben Sie:

> I use the latest Bugzilla version downloaded with "bzr co
> bzr://bzr.mozilla.org/bugzilla/trunk bugzilla" and subsequent upgrades
> with "bzr update" and successfull "./checksetup.pl".

This is a non stable development version, simply don't use it in a
production environment.

Mit freundlichen Grüßen,

Thorsten Schöning

--
Thorsten Schöning E-Mail:Thorsten....@AM-SoFT.de
AM-SoFT IT-Systeme http://www.AM-SoFT.de/

Telefon...........05151- 9468- 55
Fax...............05151- 9468- 88
Mobil..............0178-8 9468- 04

AM-SoFT GmbH IT-Systeme, Brandenburger Str. 7c, 31789 Hameln
AG Hannover HRB 207 694 - Geschäftsführer: Andreas Muchow

[Adam Ryczkowski]

On 24.02.2013 07:48, James Barwick wrote:
> Are you proxy forwarding or NAT'd through a firewall (as sometimes is the case with Virtual enterprise servers.
>
> Check any proxy forward timeous or firewall connection timeouts.
>
> Since there is nothing in you logs, I am suspicious of a proxy gateway.( or reverse NAT)
Thank you for your attention.
I heeded to the Thorsten Schöning's advice and downgraded to the 4.4rc2
version and problem "magically" vanished - it must be some kind of
reggression in the development code. But if you care, I am ready to
assist you with further troubleshooting this problem.

Bugzilla (i.e. Apache 2) connects with the outside world through the
pound proxy installed on the host which redirects http requests
according to the domain. For all other network traffic, bugzilla is
installed is behind NAT. Maybe this is a source of the confusion - but
it doesn't explain why it usually work on some guests and why it doesn't
on other. And I say usually, because the problem sometimes (although
rarely) was seen on other guests, and sometimes it didn't manifest on
the guest which is suffering the most. And when it work, there is no
noticeable lag in the submit request (contrary to the situation when the
:internal server error" occur).

Anyway, doesn't Bugzilla offer some kind of "debug mode" (like mediawiki
does) which can show a little more meaningful error
messages?<skype:sisteczko>

--
Adam

[jbarwick]

It actually makes perfect sense.

This issue is related to the proxy not getting a response in a timely fashion from the "guest".

This can be related to a variety of issues. VM latency, load, etc.etc. Each guest is different. And, I hope you don't have too many guests on a host else you could have some serious performance problems. If you don't know where your guest are physically located (which host), you could have a variety of network issues.

On the proxy, I would increase the timeout values to handle cases where the guest is slow to respond (if you have access to the proxy config).

Good luck.


On 24 Feb, 2013, at 5:27 PM, Adam Ryczkowski <adam.ry...@statystyka.net>
wrote:

> _______________________________________________
> support-bugzilla mailing list
> support-...@lists.mozilla.org
> https://lists.mozilla.org/listinfo/support-bugzilla
> PLEASE put support-...@lists.mozilla.org in the To: field when you reply.

[Adam Ryczkowski]

On 24.02.2013 10:56, jbarwick wrote:
> It actually makes perfect sense.
>
> This issue is related to the proxy not getting a response in a timely fashion from the "guest".
>
> This can be related to a variety of issues. VM latency, load, etc.etc. Each guest is different. And, I hope you don't have too many guests on a host else you could have some serious performance problems. If you don't know where your guest are physically located (which host), you could have a variety of network issues.
>
> On the proxy, I would increase the timeout values to handle cases where the guest is slow to respond (if you have access to the proxy config).
>
> Good luck.

The lag I get before being informed about the"internal server error" is
approximately 15 seconds, the same as the timeout of my pound proxy. And
I confirm, that the error message itself is produced by the pound, not
bugzilla! Thanks a lot, jbarwick!

Does it mean, that if I walk around the proxy by forwarding the port 80
from the guest into real world the problem will vanish? I guess so.

As I said before, the problem is solved with downgrade. But if anyone
here @ bugzilla is interested in troubleshooting the problem I am ready
to assist (I don't know how program in Perl though).

--
P.S.
Although the cause of the problem seems to move away from being entirely
on the bugzilla's side, I still don't understand one thing: my bug
database is small at the moment (only ca 80 bugs), and it never should
take more than half a second to process any request on such database.
And the pound have worked for us for years with the default timeout. I
think, that there is a second culprit inside bugzilla, that tries to
contact something nonexistent on the network. Maybe the problem was in
the fact, that the bugzilla was installed with different network setup
(other IP address) than it is working right now? But I did sudo
./checksetup.pl, so even if that was problem it should have been fixed.

[jbarwick]

Apache2's ProxyTimeout for ProxyPass and ProxyPassReverse in mod_proxy is about 300 seconds (This is the value of TimeOut by default).

In any case, in my application, I have to set it to at least 600 seconds. (not a bugzilla app). Oh...and setting ProxyTimeout to a value larger than TimeOut is probably not a great idea.

Don't know what the default is for the pound proxy. You should check that it's not small...15 seconds is quite small.

I run bugzilla behind a Proxy forwarding server. I'm an advocate of them. They are very helpful. They are the only way to Forward to multiple hosts by URL (not reverse NAT'ing port 80).

I haven't tried any of the later bugzilla versions (development versions). I hope there isn't something in there that we need to worry about.

Good luck.



On 24 Feb, 2013, at 6:38 PM, Adam Ryczkowski <adam.ry...@statystyka.net>
wrote:

[Adam Ryczkowski]

On 24.02.2013 12:37, James Barwick wrote:
> Hey... In your bugzilla setup, check the URL's for the graphics generators (sorry, I forget the exact setting). Make sure those URL's resolve or.. Do as I once did, install the graphics generators locally.
I just received report of the same problem when submiting edits to...
our mediawiki (on different guest, but on the same physical machine). So
I am investigating this lead, as the problem might be, after all,
completely irrelevant to the bugzilla. I suspect pound, as restarting
this service makes everything back to work again.

And, by the way, changing the timeouts from 15 to 120 seconds didn't
solve the problem.

I confirm, that bypassing pound with port redirection doesn't reproduce
this problem.

Until I am sure that the problem is bugzilla-specific I'll not post to
this forum again.

Thank you for your time and valuable suggestions.

[Thorsten Schöning]

Guten Tag Adam Ryczkowski,
am Sonntag, 24. Februar 2013 um 14:01 schrieben Sie:

> I just received report of the same problem when submiting edits to...

Sounds like a problem with sending mails to me, maybe your mail server
resolves sometimes through the proxy, sometimes not or something like
that.

[Thorsten Schöning]

Guten Tag Adam Ryczkowski,
am Sonntag, 24. Februar 2013 um 10:27 schrieben Sie:

> Anyway, doesn't Bugzilla offer some kind of "debug mode" (like mediawiki
> does) which can show a little more meaningful error
> messages?

This should still apply, but I'm not aware of other logging like
log4perl or else.

http://www.bugzilla.org/docs/2.20/html/general-advice.html

[James Barwick]

Are you proxy forwarding or NAT'd through a firewall (as sometimes is the case with Virtual enterprise servers.

Check any proxy forward timeous or firewall connection timeouts.

Since there is nothing in you logs, I am suspicious of a proxy gateway.( or reverse NAT)

Sent from my iPhone

On 24 Feb, 2013, at 10:25 AM, "Adam Ryczkowski" <adam.ry...@statystyka.net> wrote:

> On one of my seemingly identical virtual severs I always receive an error message "An internal server error occurred. Please try again later." after commiting change or posting a new bug with http user interface. The data commits just fine, but the error message is annoying and definitely doesn't look good. And commiting with Mylyn (Eclipse) is broken, since Mylyn treats this error message as error and assumes the actual data in fact is _not_ commited.

>
> There is nothing on apache2/error.log and mysql.err.
>
> I hope there is a way to make Bugzilla give more meaningfull error reports. I just couldn't find how to enable it.
>
> * * *
>

> I use the latest Bugzilla version downloaded with "bzr co bzr://bzr.mozilla.org/bugzilla/trunk bugzilla" and subsequent upgrades with "bzr update" and successfull "./checksetup.pl".
>

> Bugzilla 4.5 on standard LXC container on Ubuntu Quantal server with kernel "3.7.6-030706-generic".
>

> --
>
> Adam Ryczkowski
> +48505919892 <callto:+48505919892>
> Skype:sisteczko <skype:sisteczko>
>

[James Barwick]

Hey... In your bugzilla setup, check the URL's for the graphics generators (sorry, I forget the exact setting). Make sure those URL's resolve or.. Do as I once did, install the graphics generators locally.

Sent from my iPhone

On 24 Feb, 2013, at 6:39 PM, "Adam Ryczkowski" <adam.ry...@statystyka.net> wrote:

> On 24.02.2013 10:56, jbarwick wrote:
>> It actually makes perfect sense.
>>
>> This issue is related to the proxy not getting a response in a timely fashion from the "guest".
>>
>> This can be related to a variety of issues. VM latency, load, etc.etc. Each guest is different. And, I hope you don't have too many guests on a host else you could have some serious performance problems. If you don't know where your guest are physically located (which host), you could have a variety of network issues.
>>
>> On the proxy, I would increase the timeout values to handle cases where the guest is slow to respond (if you have access to the proxy config).
>>
>> Good luck.
> The lag I get before being informed about the"internal server error" is approximately 15 seconds, the same as the timeout of my pound proxy. And I confirm, that the error message itself is produced by the pound, not bugzilla! Thanks a lot, jbarwick!
>
> Does it mean, that if I walk around the proxy by forwarding the port 80 from the guest into real world the problem will vanish? I guess so.
>
> As I said before, the problem is solved with downgrade. But if anyone here @ bugzilla is interested in troubleshooting the problem I am ready to assist (I don't know how program in Perl though).
>
> --
> P.S.
> Although the cause of the problem seems to move away from being entirely on the bugzilla's side, I still don't understand one thing: my bug database is small at the moment (only ca 80 bugs), and it never should take more than half a second to process any request on such database. And the pound have worked for us for years with the default timeout. I think, that there is a second culprit inside bugzilla, that tries to contact something nonexistent on the network. Maybe the problem was in the fact, that the bugzilla was installed with different network setup (other IP address) than it is working right now? But I did sudo ./checksetup.pl, so even if that was problem it should have been fixed.
>