Regarding: "Probably there is no IO"
When I check "ceph -s", it says there is activity. Not much, but some
activity:
io:
client: 316 KiB/s rd, 10 KiB/s wr, 1 op/s rd, 1 op/s wr
And no activity doesn't explain why cephfs-top complains about not finding
the cluster.
Re: using the options available:
# cephfs-top --cluster ceph
cluster ceph does not exist
# cephfs-top --id 767edb64-4552-43a2-a128-34d693343904
cluster ceph does not exist
# cephfs-top --conffile /etc/ceph/ceph.conf
cluster ceph does not exist
It really doesn't want to connect to the cluster! ;-)
In the mgr logfile, I see an issue that might be related ( I never checked
the mgr log, as all seemed to be working fine) showing every second or so:
1 mgr finish mon failed to return metadata for mds.cph-a-002: (2) No such
file or directory
This actually seems to be the same issue as this user experienced:
http://lists.ceph.com/pipermail/ceph-users-ceph.com/2018-April/026126.html
The active mds gives the same error if I do a "ceph mds metadata <mds>":
# ceph mds metadata cph-a-002
{}
Error ENOENT:
The secondary works:
# ceph mds metadata cph-h-001
{
"addr": "[v2:10.0.1.101:6800/1267106302,v1:10.0.1.101:6801/1267106302]",
"arch": "x86_64",
"ceph_release": "pacific",
"ceph_version": "ceph version 16.2.5
(0883bdea7337b95e4b611c768c0279868462204a) pacific (stable)",
"ceph_version_short": "16.2.5",
....etc
If I reboot the mgr, I see some dubious messages:
0 ms_deliver_dispatch: unhandled message 0x55b3d8c449a0 mon_map magic: 0 v1
from mon.1 v2:10.0.1.102:3300/0
...
-1 client.0 error registering admin socket command: (17) File exists
Then it continues with the error message above every second or so.
I then reboot the primary mon, which shows some iteresting messages:
3 rocksdb: [le/block_based/filter_policy.cc:584] Using legacy Bloom filter
with high (20) bits/key. Dramatic filter space and/or accuracy improvement
is available with format_version>=5.
...
0 mon.cph-h-001@-1(???).osd e35181 crush map has features
3314933069573799936, adjusting msgr requires
...
1 mon.cph-h-001@0(electing) e26 collect_metadata : no unique device id for
: fallback method has no model nor serial'
In the active mds logs, not much is happening.
Hope this gives a better overview of what's (not) happening.
_______________________________________________
ceph-users mailing list -- ceph-users@xxxxxxx
To unsubscribe send an email to ceph-users-leave@xxxxxxx
