I recently installed zabbix on my mythtv box to monitor it remotely, both agent and server running on the same box, and a day or two later I noticed that the PC had hung up; I could still move windows around on the screen, but nothing else worked and I couldn't log in with ssh. I rebooted it and looked at the zabbix history and it showed the load average rolling along around 4-6 for a few hours while it was transcoding video files (the machine can run four threads, so that's OK), and then an exponential rise to 20 or so before the history log ended as the system froze up.
After rebooting I left top running to see if I could find out what caused that if it happened again, and it locked up the next night too, again while transcoding some video files. Checking the system I could see that top had been frozen for about half an hour, and when it froze zabbix_agentd was at the top of the list using 100% CPU and the load average was over 16.
I'm running the 1.4.6 version from Ubuntu 8.10, and while I've found a few similar bugs mentioned here on the forum they seem to apply to earlier releases; does anyone know what might be causing this? I can't connect to the agent process to try to see what it's doing because I can't do anything on the system other than move windows around or watch the clock update.
About all I can think of is that the CPU may be so heavily overloaded that the agent can't keep up with the requests from the server and eventually overloads the CPU so much that it can't achieve anything. I have a number of UserParameters configured that grep data out of system files (e.g. /proc/diskstat), but nothing that should take long to run.
After rebooting I left top running to see if I could find out what caused that if it happened again, and it locked up the next night too, again while transcoding some video files. Checking the system I could see that top had been frozen for about half an hour, and when it froze zabbix_agentd was at the top of the list using 100% CPU and the load average was over 16.
I'm running the 1.4.6 version from Ubuntu 8.10, and while I've found a few similar bugs mentioned here on the forum they seem to apply to earlier releases; does anyone know what might be causing this? I can't connect to the agent process to try to see what it's doing because I can't do anything on the system other than move windows around or watch the clock update.
About all I can think of is that the CPU may be so heavily overloaded that the agent can't keep up with the requests from the server and eventually overloads the CPU so much that it can't achieve anything. I have a number of UserParameters configured that grep data out of system files (e.g. /proc/diskstat), but nothing that should take long to run.
Comment