No output from Gluster statedump 'kill -USR1'
Issue
If a glusterfs process is using a lot of memory on a gluster client host, customers are often asked to create a statedump of the process with the comand kill -USR1 pid-of-glusterfs-process. The output of that command is supposed to be written to /var/run/gluster. However, if that directory does not already exist, the statedump produces no output.
Environment
A RHEL host mounting a gluster volume:
[root@client01 ~]# mount -t glusterfs gluster01:/vol03 /mnt/vol03
The mount is managed by a glusterfs process:
[root@client01 ~]# ps -ef | grep glusterfs | grep "/mnt/vol03"
root 10914 1 0 10:55 ? 00:00:00 /usr/sbin/glusterfs --volfile-server=gluster01 --volfile-id=/vol03 /mnt/vol03
Support may ask the customer to create a statedump of that process:
[root@client01 ~]# kill -USR1 10914
Support asks the customer to gather the output from the /var/run/gluster directory:
# ls -l /var/run/gluster
Total ...
-rw-------. 1 root root 39406 Jun 25 11:00 glusterdump.10914.dump.1561474824
On a gluster node running the glusterd process, the /var/tmp/gluster directory is created automatically. On a client, it is not. If the directory does not already exist, the statedump creates no output
[root@client01 ~] ls -l /var/tmp/gluster
ls: cannot access /var/tmp/gluster: No such file or directory
Subscriber exclusive content
A Red Hat subscription provides unlimited access to our knowledgebase, tools, and much more.
