After Update to SVN 4928 web pages time out constantly

[Tom Elliott]

Please update and try again, I’m trying to limit the number of sessions.

[Joseph Hales]

updating here is lscpu output

Architecture: x86_64
CPU op-mode(s): 32-bit, 64-bit
Byte Order: Little Endian
CPU(s): 16
On-line CPU(s) list: 0-15
Thread(s) per core: 2
Core(s) per socket: 4
Socket(s): 2
NUMA node(s): 2
Vendor ID: GenuineIntel
CPU family: 6
Model: 26
Stepping: 5
CPU MHz: 2261.117
BogoMIPS: 4521.87
Virtualization: VT-x
L1d cache: 32K
L1i cache: 32K
L2 cache: 256K
L3 cache: 8192K
NUMA node0 CPU(s): 0,2,4,6,8,10,12,14
NUMA node1 CPU(s): 1,3,5,7,9,11,13,15

[Joseph Hales]

Old errors reoccur plus some new ones.

[Tue Oct 13 11:32:00.141558 2015] [:error] [pid 3576] [client 10.110.50.40:65405] PHP Warning:  mysqli::query(): Couldn't fetch mysqli in /var/www/html/fog/lib/db/MySQL.class.php on line 62
[Tue Oct 13 11:32:01.035167 2015] [:error] [pid 3579] [client 10.117.50.11:53325] PHP Fatal error:  Uncaught exception 'ReflectionException' with message 'Class EventManager does not exist' in /var/www/html/fog/lib/fog/FOGBase.class.php:64\nStack trace:\n#0 /var/www/html/fog/lib/fog/FOGBase.class.php(64): ReflectionClass->__construct('EventManager')\n#1 /var/www/html/fog/commons/init.php(196): FOGBase->getClass('EventManager')\n#2 /var/www/html/fog/commons/base.inc.php(3): require_once('/var/www/html/f...')\n#3 /var/www/html/fog/service/servicemodule-active.php(2): require_once('/var/www/html/f...')\n#4 {main}\n  thrown in /var/www/html/fog/lib/fog/FOGBase.class.php on line 64
[Tue Oct 13 11:32:01.089682 2015] [:error] [pid 3667] [client 10.1.50.220:53541] PHP Fatal error:  Uncaught exception 'ReflectionException' with message 'Class ServiceManager does not exist' in /var/www/html/fog/lib/fog/FOGBase.class.php:64\nStack trace:\n#0 /var/www/html/fog/lib/fog/FOGBase.class.php(64): ReflectionClass->__construct('ServiceManager')\n#1 /var/www/html/fog/lib/fog/FOGCore.class.php(14): FOGBase->getClass('ServiceManager')\n#2 /var/www/html/fog/lib/fog/FOGCore.class.php(217): FOGCore->getSetting('FOG_THEME')\n#3 /var/www/html/fog/commons/init.php(192): FOGCore->setSessionEnv()\n#4 /var/www/html/fog/commons/base.inc.php(3): require_once('/var/www/html/f...')\n#5 /var/www/html/fog/service/servicemodule-active.php(2): require_once('/var/www/html/f...')\n#6 {main}\n  thrown in /var/www/html/fog/lib/fog/FOGBase.class.php on line 64
[Tue Oct 13 11:32:01.230274 2015] [:error] [pid 5538] [client 10.113.50.91:50426] PHP Fatal error:  Uncaught exception 'ReflectionException' with message 'Class ServiceManager does not exist' in /var/www/html/fog/lib/fog/FOGBase.class.php:64\nStack trace:\n#0 /var/www/html/fog/lib/fog/FOGBase.class.php(64): ReflectionClass->__construct('ServiceManager')\n#1 /var/www/html/fog/lib/fog/FOGCore.class.php(14): FOGBase->getClass('ServiceManager')\n#2 /var/www/html/fog/lib/fog/FOGCore.class.php(217): FOGCore->getSetting('FOG_THEME')\n#3 /var/www/html/fog/commons/init.php(192): FOGCore->setSessionEnv()\n#4 /var/www/html/fog/commons/base.inc.php(3): require_once('/var/www/html/f...')\n#5 /var/www/html/fog/service/snapins.checkin.php(2): require_once('/var/www/html/f...')\n#6 {main}\n  thrown in /var/www/html/fog/lib/fog/FOGBase.class.php on line 64
[Tue Oct 13 11:34:49.125571 2015] [mpm_prefork:error] [pid 5892] AH00161: server reached MaxRequestWorkers setting, consider raising the MaxRequestWorkers setting```

[Joseph Hales]

had to delete some of the errors as the post kept getting marked as spam by the filter but they seem to be repeats.

[Wayne Workman]

@Joseph-Hales said:

had to delete some of the errors as the post kept getting marked as spam by the filter but they seem to be repeats.

You can get around that by uploading the logs in a .txt file.

[Joseph Hales]

Ok yet another sign FOG hates me I come back from lunch my forum session wouldn’t let me log back in kept saying forbidden until I closed the window and reopened it

[Joseph Hales]

Should I roll back to SVN 4700 and if so what check out version is that?

[Wayne Workman]

@Joseph-Hales if you go to your svn/trunk directory you can run svn info to get the revision number that you currently have checked out.

You know the only safe method of downgrading is with a snapshot, right?

[Joseph Hales]

Any other suggestions it seems like after a restart sql hits some sort of loop state and just locks up with that apache maxworkers error.

[Wayne Workman]

@Joseph-Hales If it’s possible, disable the interface on the fog server at boot time.

On CentOS 7 / RHEL 7 / Fedora 19-22 you may do this by editing /etc/sysconfig/network-scripts/<interface name> and changing the line ONBOOT="yes" to ONBOOT="no"

http://ask.xmodulo.com/configure-static-ip-address-centos7.html

Then reboot and see how MySQL behaves… does it enter into a loop-like state then?

If not, it probably really is load related - and hopefully @Tom-Elliott can figure out what the difference is between then and now.

If it still has this problem with the interface turned off, then the problem resides on the fog server somewhere and is not load related.

[Tom Elliott]

I’ve completely rewritten session handling in the hopes that this would be fixed. I suspect that what’s happening is the sessions are being started and started many times. So after initial startup all the connections are in use for the new start up.

[Joseph Hales]

I have increased MaxConnections and MaxRequestWorkers both to 1000 installed SVN 4956 and restarted the server with the exact same results now on the attached apache error log I notice it gets to the MaxRequestWorkers suggestion and then stops untill I restart the server. This behavior also appears in my earlier posts I just didn’t note it.
Apache.log.txt

[Joseph Hales]

With the new settings I am occasionally able to get web interface pages to load 1 time out of 10 with several minutes to load each page but it doesn’t appear that Apache is logging anything anymore after it crashes on the max workers suggestion.

[Joseph Hales]

Fog ping hosts seems to be my next resource hog after mysql is this normal?

TOP 10 MEM (ps auxww --sort=-rss|head -n10)
USER       PID %CPU %MEM    VSZ   RSS TTY      STAT START   TIME COMMAND
mysql     1419  143  0.9 8723248 235708 ?      Ssl  10:14 252:42 /usr/sbin/mysqld
root      1956 32.5  0.5 312728 129040 ?       S    10:14  57:12 /usr/bin/php -q /opt/fog/service/FOGPingHosts/FOGPingHosts
root      2027  0.0  0.1  90772 25100 ?        Ss   10:14   0:00 /usr/bin/perl /usr/share/webmin/miniserv.pl /etc/webmin/miniserv.conf
root      2174  0.0  0.1 800960 24888 ?        Ssl  10:14   0:01 /opt/simpana/Base/cvd
sealion   1471  0.3  0.0 809564 22208 ?        Sl   10:14   0:38 python /usr/local/sealion-agent/etc/init.d/../../bin/sealion.py start
www-data  2695  0.3  0.0 328436 20700 ?        S    10:15   0:36 /usr/sbin/apache2 -k start
www-data  2359  0.2  0.0 329392 20344 ?        S    10:14   0:28 /usr/sbin/apache2 -k start
www-data  2702  0.2  0.0 402536 18924 ?        S    10:15   0:30 /usr/sbin/apache2 -k start
www-data  2299  0.2  0.0 475844 18440 ?        S    10:14   0:30 /usr/sbin/apache2 -k start

Mod edited to use code box.

[Joseph Hales]

Here is what a fog client sees when the server Apache log is spamming the following.

fog.log

[Wed Oct 14 14:38:20.061493 2015] [:error] [pid 2706] [client 10.109.49.82:50064] PHP Warning:  mysqli::query(): Couldn't fetch mysqli in /var/www/html/fog/lib/db/MySQL.class.php on line 62

[Sebastian Roth]

Maybe your database is kind of crashed again?? https://forums.fogproject.org/topic/5764/fog-web-interface-times-out-loading-almost-any-page

Have you ever looked into mysql error logging?? May be start here: http://ronaldbradford.com/blog/monitoring-mysql-the-error-log-2009-09-16/ and http://www.pontikis.net/blog/how-and-when-to-enable-mysql-logs

If I got this right mysql should log errors to syslog in standard configuration. Not sure about that on all linux systems (RedHat vs. Debian vs. Arch vs. ???). But give it a try zgrep mysql /var/log/syslog* (output might be quite long!)

[Tom Elliott]

@Joseph-Hales ping hosts could be a culprit in that it pings all of your hosts as a service separately from the main. The ping time happens every five minutes after the last completion. My guess is this IS the sole reason to the problem you’re seeing considering from what I recall you have about 6k hosts. You can disable the service and my guess is you’ll see much better stays overall on your server, including MySQL.

[Tom Elliott]

More specifically.

FOGPingHosts works as a service similar to FOGMulticastManager or FOGImageReplicator. It’s purpose is to ping the hosts and report the “status” so that you’re not waiting on the List All/Search on the GUI to move between pages.

It pings each host and updates the ping status column on the database. This iterates every 5 minutes after the last one updates. So every 5 minutes it’s pinging all hosts of the server.

Now I’ve added a usleep of 1/1000th of a second per each host ping to allow the CPU to cool down between themselves so we don’t completely overload your CPU. I’ll up this value just to help slow things down a bit to 1/10th of a second. I’m not going to adjust the sleep time as 5 minutes is okay for most, though you can adjust your value in the Config.class.php.

This would explain the mysql usage AND the FOGPingHosts taking up most of the time.

[Joseph Hales]

I updated to the latest SVN same issue after restart where would I find Config.Class.php and what would I edit to remove ping hosts entirely?

[Tom Elliott]

@Joseph-Hales what version? Do you have ping on?