- Plesk 12.5 for Linux
- Plesk Onyx for Linux
- Plesk 11.x for Linux
- Plesk 12.0 for Linux
Symptoms vary, but only occur in Virtuozzo containers (former Parallels Virtuozzo Container - PVC) or OpenVZ: Plesk or system services fail to start, malfunction, or even crash.
Not enough resources are allocated to the container.
Note: This article only deals with the UBC resource management model in Virtuozzo containers. For early PVC versions (including 3.0-4.6) containers may be running in SLM mode or even both modes. In these cases, it is required to check SLM memory usage. This is best completed via the Virtual Automation (former Parallels Virtual Automation - PVA) web interface. More information on SLM can be found in the PVC documentation.
As an example, let's imagine that Apache fails to start on the container.
To verify that the issue occurs due to resource shortage:
Log in to the container via SSH and issue the following command to record the current state of User Beancounters:
# cat /proc/user_beancounters > before_failure.txt
Reproduce the issue (try starting Apache)
Issue this command to record the new state of UBC:
# cat /proc/user_beancounters > after_failure.txt
Open the two files (
after_failure.txt) and check the values in the '
failcnt' column. If any of increased value indicates that the particular resource shortage occurred. If so, follow the resolution section below.
Increase the corresponding UBC limits for the container by PVA means or run the following command on the Hardware Node:
# vzctl set <CT_ID> --save --<RESOURCE_NAME> <BARRIER>:<LIMIT>
- is container ID
- is the name of the resource (can be taken from the user_beancounters file)
- is the new barrier (soft limit) for the resource (check the current value in the user_beancounters file)
- is the new limit (hard limit) for the resource (check the current value in the user_beancounters file)