Issue with upgrade

[jtata]

I am experiencing a major issue with my Nagios XI install. After performing an update to php from yum I started getting "exit signal segmentation fault" errors. Doing some research I found this was corrected in a later version of Nagios, so I began the upgrade process.

The upgrade script has been stuck on the following step:

Patching NDOUtils...
Applying NDOUtils patch level 101
ALTER TABLE `nagios_logentries` ADD INDEX ( `logentry_time` ) ;

After about 2 hours I was forced to kill it and revert my VM to a snapshot. Any advice how to resolve?

[mguthrie]

What version of XI were you upgrading from?

Is your mysql database offloaded to a second server?

Lets also run the DB maintenance script and make sure there's no table corruption there.

Code: Select all

cd /usr/local/nagiosxi/cron
rm -f ../var/dbmaint.lock
./dbmaint.php

If you see any tables listed as being crashed, run the Db repair procedure below.
http://assets.nagios.com/downloads/nagi ... irdatabase

[jtata]

Version is 2009R1.3G, which I realize is very out of date. MySQL is installed on the same server. No crashed tables found during DB maintenance.

[jtata]

It strikes me an easier option might be to download the newest VM image and simply restore a backup. Will this work? Can a backup of 2009 be restored to 2011?

[mguthrie]

It's also possible that your logentries table is very large, so the table alteration was timing out. There were some changes made to the ndoutils table from 2009 to 2011, but the backup and restore might work. I'd say it's worth a try as long as you've got a VM where you can revert the snapshot.

[jtata]

I reran the upgrade script after truncating the logentries table. It completed successfully, however I still have my original problem: accessing the web interface repeatedly fails with "Unable to load the webpage because the server sent no data." Reloading the page a few times will bring the correct data. /etc/httpd/logs/error_log contains the following:

Code: Select all

[notice] child pid 3592 exit signal Segmentation fault (11)

This occurs every few seconds with a different PID.

Suggestions?

[mguthrie]

Yuck. Seg faults in apache are not easy to trace.

Can you send us your Linux distro information that the install is on, as well as the output from the below command:

Code: Select all

yum list installed | grep php

[jtata]

I think I may have resolved the issue. I ran pvp -v and found an error about sourceguardian. I noticed this link in the FAQ:
http://support.nagios.com/wiki/index.ph ... ian_Errors

So I added the line to php.ini and restarted Apache. It seems to have corrected the problem. Nagios is loading correctly and I do not see any segfault errors anymore.

[mguthrie]

Good deal, I was sweatin' on that one a bit ; )