Webinator Hangs - Server

joseph.gresham · Post by **joseph.gresham** » Mon Jun 12, 2006 6:00 am

Hi

We are currently experiencing a strange issue with our installation of Webinator. The problem that we a experiencing is the once the admin interface has loaded any attempts to access the left menu items completely hangs the server with a reset of Apache usually the only way to fix the problem. Also when Webinator hangs it also hangs the website which is running on LiveLink.

Any help you could give on this would be greatly appreciated.

Server details

OS Red Hat Enterprise 4
Webinator 5.1 - Linux - 2-4-libc2.2-X86
I have also attached some error logs which may be of some use.

LOGS

*******************************************

In livelink dbengine.log I can see a series of these errors

Jun 09 2006 10:33:25 pid:00013339 th:3049704368: [9004] broken communication channel: Releasing all connects of [DSN=wpel;UID=nat_
equip_live1]
Jun 09 2006 10:33:25 pid:00013339 th:3049704368: [9034] General SQL error
[WCM Server][Section: 2; Top URL: eqd2.hants.gov.uk/index/equipment/productdirectory/findaproductsearchresults]
ORA-03113: end-of-file on communication channel

Jun 09 2006 10:33:32 pid:00013138 th:3039214512: [9034] General SQL error
[WCM Server][Section: 2; Top URL: eqd2.hants.gov.uk/save_page.gif]
ORA-03135: connection lost contact

Jun 09 2006 10:33:32 pid:00013138 th:3028724656: [9004] broken communication channel: Releasing all connects of [DSN=wpel;UID=nat_
equip_live1]
Jun 09 2006 10:33:32 pid:00013138 th:3028724656: [9034] General SQL error
[WCM Server][Section: 2; Top URL: eqd2.hants.gov.uk/save_page.gif]
ORA-03114: not connected to ORACLE

******************************************

In apache error.log
[Fri Jun 09 10:35:11 2006] [error] [client 192.168.14.12] Premature end of script headers: texis, referer: http://eqd2.hants.gov.uk/cgi-bin/texis/ ... anced.html

******************************************

In monitor.log
200 2006-06-09 10:33:06 (13410) Database Monitor on /usr/local/morph3/texis/eqd/db1/ starting 200 2006-06-09 10:39:06 (13410) Database Monitor on /usr/local/morph3/texis/eqd/db1/ exiting

******************************************

Post by **mark** » Mon Jun 12, 2006 11:14 am

Never heard of anything like that. Do you see any texis process running long and/or using a lot of cpu after this happens?

None of the db errors are related to webinator.
Check vortex.log for webinator errors.

joseph.gresham · Post by **joseph.gresham** » Tue Jun 13, 2006 11:31 am

Unfortunately the vortex.log seems to be empty except for the following line.

100 2006-06-08 15:54:46 /webinator/dowalk:9565: User PUBLIC has been added without a password.
015 2006-06-12 11:36:22 Unknown option `--version' (-h for help)

Also, there seems to no indication of the texis process consuming large amounts of cpu or memory resources. If I could get an official ticket number for the ticket I sent in yeserday (with the exact content ) I'll get the guys who actually host the servers to take up the call as this is something that we need to solve quite urgently.

Regards

Post by **mark** » Tue Jun 13, 2006 1:55 pm

If you sent the same thing to the board and as a ticket the ticket probably got tagged as a dup. I'll see about resurrecting it. Please send questions to one place or the other rather than both.

Maybe texis is incorrectly installed so that it doesn't have permissions to write to the log file and other things. Make sure texis is setuid to the proper user and that the files are owned by that user and that apache isn't suppressing the setuid bit or is running texis as the proper user when run as a CGI.

What version and platform of apache? Nothing a CGI program does should be able to take out the whole http server.

joseph.gresham · Post by **joseph.gresham** » Thu Jun 15, 2006 4:40 am

Texis was installed as the root user but set to run as the Livelink user. In the running processed list it shows that it is running as the Livelink user - see entry from process list below

livelink 15864 1 0 Jun01 ? 00:00:00 monitor -d /usr/local/morph3/texis/testdb/ -z

The log files are owned by this user and are being written to - even the vortex.log. Although it wasn't written to yesterday during testing, it has been written to in the past on two separate occasions. See entries below:

100 2006-06-08 15:54:46 /webinator/dowalk:9565: User PUBLIC has been added without a password.

015 2006-06-12 11:36:22 Unknown option `--version' (-h for help)

We are running Apache 2.0.52

And the operating system is Red hat Enterprise release 4.

Post by **John** » Thu Jun 15, 2006 6:55 am

Should apache be running texis directly, or will LiveLink be involved? I'm not quite clear on what the flow should be between the browser, Apache, LiveLink and Texis.

johann.richard · Post by **johann.richard** » Thu Jul 13, 2006 3:24 pm

Hello

We experience the very same problem, that is, we have texis/monitor running as apache user in cgi-bin and have the same hangs. I opened a ticket today, but was wondering if you finally got solution here?

Thanks

johann.richard · Post by **johann.richard** » Fri Jul 14, 2006 1:32 pm

Hi Joseph

We have the exact same problems as you had. Thunderstone Support did not provide any answers, so I was wondering if (and possibly how) you could solve the problem?

I would very much appreciate any hints here.

Best regards
Johann

joseph.gresham · Post by **joseph.gresham** » Mon Jul 17, 2006 6:59 am

Hi

I'm afraid that we has still not found a solution to this problem. The build team are currently planning on re-installing Webinator on a new, clean, server to see if this solves the problem. We are making the assumption (and its a rather large assumption) that there may be some sort of incompatibility between Webinator and some of the other products on the existing server.