I had this weekend a process stuck in D state writing to a cephfs kernel
mount, causing the load of the server go to 80 (normally around 1).
Forcing me to reboot it.
I think this problem is related to the networking between this vm and
ceph nodes. Rsync also sometimes complains about a broken pipe.
I am not updating the kernels to often on this server, because I do not
want to reboot it if it is not necessary. I am still at
3.10.0-514.21.2.el7.x86_64.
Now I am going to upgrade of course to the latest of CentOS 7.6.
Regardless of investigating the networking issue. Are there any other
mitigations I can apply.
- Maybe move to the 4.x kernel, does that make a big difference?
- how to prevent the D state process to accumulate so much load?
Thanks,
_______________________________________________
ceph-users mailing list
ceph-users@lists.ceph.com
http://lists.ceph.com/listinfo.cgi/ceph-users-ceph.com