mon_max_pg_per_osd ignored after restarting daemons
We have a Mimic 13.2.8 cluster that is showing WARN: "too many PGs per OSD (239 > max 200)" However, I changed the mon_max_pg_per_osd level to 300 and restarted all of the MON and OSD daemons, but the warning persists. If I check the settings using the admin-daemon socket, they all report the new value (300), but the warning will not go away and still reports it as 200 in the health status. How can I clear this state and get the system to reflect the correct value?
Hi, this is actually reported by MRG daemons, so you have to set it in ceph.conf on nodes where the MGR are running under [global] or [mgr] section and restart the mgr services. Then the warning should go off. Kind Regards, Tomas Petr Senior Technical Support Engineer Red Hat Ceph Storage On Sun, Jan 19, 2020 at 3:57 PM Wyllys Ingersoll < wyllys.ingersoll@keepertech.com> wrote:
We have a Mimic 13.2.8 cluster that is showing WARN: "too many PGs per OSD (239 > max 200)"
However, I changed the mon_max_pg_per_osd level to 300 and restarted all of the MON and OSD daemons, but the warning persists.
If I check the settings using the admin-daemon socket, they all report the new value (300), but the warning will not go away and still reports it as 200 in the health status.
How can I clear this state and get the system to reflect the correct value?
_______________________________________________ Dev mailing list -- dev@ceph.io To unsubscribe send an email to dev-leave@ceph.io
thanks, that fixed it! I should have thought of that sooner. All of the searches I did for the issue said to restart the mon and osd services, but those were probably written pre-Mimic. On Sun, Jan 19, 2020 at 10:37 AM Tomas Petr <tpetr@redhat.com> wrote:
Hi,
this is actually reported by MRG daemons, so you have to set it in ceph.conf on nodes where the MGR are running under [global] or [mgr] section and restart the mgr services. Then the warning should go off.
Kind Regards,
Tomas Petr Senior Technical Support Engineer Red Hat Ceph Storage
On Sun, Jan 19, 2020 at 3:57 PM Wyllys Ingersoll < wyllys.ingersoll@keepertech.com> wrote:
We have a Mimic 13.2.8 cluster that is showing WARN: "too many PGs per OSD (239 > max 200)"
However, I changed the mon_max_pg_per_osd level to 300 and restarted all of the MON and OSD daemons, but the warning persists.
If I check the settings using the admin-daemon socket, they all report the new value (300), but the warning will not go away and still reports it as 200 in the health status.
How can I clear this state and get the system to reflect the correct value?
_______________________________________________ Dev mailing list -- dev@ceph.io To unsubscribe send an email to dev-leave@ceph.io
participants (2)
-
Tomas Petr
-
Wyllys Ingersoll