Processes building up on server running cacti

[HaIrLeSsHoBo]

I have noticed over the past few weeks that processes appear to be stacking up on my server that is running Cacti. For example:

Code: Select all

username     17248  0.0  0.1  13016  3644 ?        S    Apr12   0:00 /usr/bin/php -q /home/username/public_html/cacti/plugins/mactrack/mactrack_scanner.php -id=2
username      5442  0.0  0.1  13016  3648 ?        S    Apr13   0:00 /usr/bin/php -q /home/username/public_html/cacti/plugins/mactrack/mactrack_scanner.php -id=2
username     16194  0.0  0.0   2788  1196 ?        S    Apr16   0:00 sh -c ping -c 1 wirelesslg1.example.org | grep icmp_seq | grep time
username     16195  0.0  0.0   4104  1388 ?        S    Apr16   0:00  \_ ping -c 1 wirelesslg1.example.org
username     16196  0.0  0.0   1828   504 ?        S    Apr16   0:00  \_ grep icmp_seq
username     16197  0.0  0.0   1828   504 ?        S    Apr16   0:00  \_ grep time
username      4011  0.0  0.0   2788  1192 ?        S    Apr16   0:00 sh -c ping -c 1 hostname.example.org | grep icmp_seq | grep time
username      4012  0.0  0.0   4104  1396 ?        S    Apr16   0:00  \_ ping -c 1 hostname.example.org
username      4013  0.0  0.0   1828   504 ?        S    Apr16   0:00  \_ grep icmp_seq
username      4014  0.0  0.0   1828   504 ?        S    Apr16   0:00  \_ grep time
username     12523  0.0  0.0   2788  1192 ?        S    Apr17   0:00 sh -c ping -c 1 wirelesslg1.example.org | grep icmp_seq | grep time
username     12524  0.0  0.0   4104  1392 ?        S    Apr17   0:00  \_ ping -c 1 wirelesslg1.example.org
username     12525  0.0  0.0   1828   504 ?        S    Apr17   0:00  \_ grep icmp_seq
username     12526  0.0  0.0   1828   500 ?        S    Apr17   0:00  \_ grep time
username     20859  0.0  0.0   2788  1196 ?        S    Apr17   0:00 sh -c ping -c 1 wirelesslg1.example.org | grep icmp_seq | grep time
username     20860  0.0  0.0   4104  1396 ?        S    Apr17   0:00  \_ ping -c 1 wirelesslg1.example.org
username     20861  0.0  0.0   1828   504 ?        S    Apr17   0:00  \_ grep icmp_seq
username     20862  0.0  0.0   1828   504 ?        S    Apr17   0:00  \_ grep time
username     23202  0.0  0.0   2788  1188 ?        S    04:25   0:00 sh -c ping -c 1 wirelessbl1.example.org | grep icmp_seq | grep time
username     23203  0.0  0.0   4104  1396 ?        S    04:25   0:00  \_ ping -c 1 wirelessbl1.example.org
username     23207  0.0  0.0   1828   508 ?        S    04:25   0:00  \_ grep icmp_seq
username     23209  0.0  0.0   1828   500 ?        S    04:25   0:00  \_ grep time
username     12469  0.0  0.1  13016  3652 ?        S    10:15   0:00 /usr/bin/php -q /home/username/public_html/cacti/plugins/mactrack/mactrack_scanner.php -id=2

Just last week (either Thursday or Friday) I killed all running processes under the username in question, and now they are stacked up again. I do not know exactly when this started, but I feel that it is the cause of my server getting overloaded after running for months unattended. Looking at my cacti graph, my server went from running around 150 processes to over 300 processes in a matter of a couple weeks. I have been monitoring my process tree and it appears that this is the only thing on the system that is building up over time.

System Information:
Slackware Linux 12.2
Linux 2.6.27.7-smp
Intel P4 Xeon 2.2GHz w/ HT
2GB RAM

Cacti 0.8.7g

Plugins:
Mactrack 2.9
Aggregate 0.75
Devices 0.4
Spikekill 1.2
Thold 0.4.1

I do not know what would be causing these processes to be stacking up like they are. Does anyone have any ideas of where to start looking?

Thanks in advance!
-Steve

[gandalf]

This looks like an old ping.pl script. We fixed it recently (unless I was not aware of this type of issues). You may download it from svn.cacti.net
R.

[edit] should have been ping.pl

[HaIrLeSsHoBo]

I've looked all through the repository and I have not been able to find a snmp.pl script. Are you certain it is Perl script and not a PHP script instead? Or perhaps a different name?

Thanks

[HaIrLeSsHoBo]

I also forgot to mention that I am currently using Spine as my polling method.

[gandalf]

I meant ping.pl
R.

[HaIrLeSsHoBo]

I checked the diff between the old version of ping.pl and the new version, and the only change is the regex pattern used for grepping icmp_[s|r]eq.. unfortunately, I am running a slightly older version of Linux that still used the icmp_seq style, so therefore that could not be the issue. Thanks for the suggestion though.

I have just checked again and in a matter of less than 24 hours I have built up two more stuck ping processes, which is a total of 8 processes. For testing purposes, I even disabled using mactrack temporarily just to see if that was causing the problem. I obviously do not get any more stuck mactrack processes, but I am still getting stuck ping processes. Any more pointers would be greatly appreciated!!

Thanks in advance!