Applies to:
Centrify Server Suite 2016 and later
Question:
The oom_adj (Out Of Memory) score is used by the operating system when the machine is low on memory. The score determines the risk involved in terminating a process to free up memory. A higher score means that the process is MORE likely to be terminated.
When a shell in opened through a PuTTY login to a Centrified CentOS 6.2 machine, the resulting oom_adj of that process is -17. The cause for concern is that the -17 would mean that the process would not be terminated. If there are say 100 users that are using a process that cannot be terminated, and the system might go into panicked mode.
To determine the current oom_adj score
1) Ensure Centrify OpenSSH is running
2) Launch PuTTY and connect to the CentOS machine
3) When the shell opens, find the process ID by entering this command:
$ echo $$
13772
4) Change to the process directory for this specific shell and check the oom_adj file
$ cd /proc/13772 (where 13772 is the Proc id from the echo in step 3)
$ cat oom_adj
5) Observe that the score is -17. For comparison, repeat these steps when stock sshd is running and observe that the score is 0.
How can the oom_adj score be modified to 0 so it is the same value used by stock SSHD?
Answer:
1) Logon to the Centrified machine as root and edit the file /etc/init/centrify-sshd.conf .
2) Find the line which reads 'oom never' and place a hash mark at the beginning of the line, as in the image below:
3) Save the file and then restart Centrify SSHD
# /etc/init.d/centrify-sshd restart