Hi,
Finally, I got a running server with files /sys/kernel/debug/ceph/xxx/
[root@docker27 525c4413-7a08-40ca-9a98-0a6df009025b.client213522]# cat mdsc
[root@docker27 525c4413-7a08-40ca-9a98-0a6df009025b.client213522]# cat monc
have monmap 2 want 3+
have osdmap 4545 want 4546
have fsmap.user 0
have mdsmap 335 want 336+
fs_cluster_id -1
[root@docker27 525c4413-7a08-40ca-9a98-0a6df009025b.client213522]# cat osdc
REQUESTS 6 homeless 0
82580 osd10 1.7f9ddac7 [10,13]/10 [10,13]/10 10000053a04.00000000 0x400024 1 write
81019 osd11 1.184ed679 [11,7]/11 [11,7]/11 1000005397b.00000000 0x400024 1 write
81012 osd12 1.cd98ed57 [12,9]/12 [12,9]/12 10000053971.00000000 0x400024 1 write,startsync
82589 osd12 1.7cd5405a [12,8]/12 [12,8]/12 10000053a13.00000000 0x400024 1 write,startsync
80972 osd13 1.91886156 [13,4]/13 [13,4]/13 10000053939.00000000 0x400024 1 write
81035 osd13 1.ac5ccb56 [13,4]/13 [13,4]/13 10000053997.00000000 0x400024 1 write
The cluster claims nothing, and shows HEALTH_OK still.
What I did is just vim a file storing on cephfs, and then it hung there. And I got a process with 'D' stat.
By the way, the whole mount directory is still in use and with no error.
What can I do to fix it?
Burkhard Linke <Burkhard.Linke@xxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxx> 于2018年8月9日周四 下午9:42写道：
Hi,
On 08/09/2018 03:21 PM, Yan, Zheng wrote:
> try 'mount -f', recent kernel should handle 'mount -f' pretty well
> On Wed, Aug 8, 2018 at 10:46 PM Zhenshi Zhou <deaderzzs@xxxxxxxxx> wrote:
>> Hi,
>> Is there any other way excpet rebooting the server when the client hangs?
>> If the server is in production environment, I can't restart it everytime.
One method that worked for me today:
- failover to other mds (during failover the client disconnect should be
denied, e.g. 'mds.ceph-storage-01 [INF] denied reconnect attempt (mds is
up:reconnect) from client.19660826 192.168.2.92:0/2522971681 (session is
closed)')
- failover to first mds again (this time, the client should not try to
connect. mds_sessions in the debug directory should not list a session)
- accessing the mountpoint triggered a reconned to the now active mds
Regards,
Burkhard
_______________________________________________
ceph-users mailing list
ceph-users@xxxxxxxxxxxxxx
http://lists.ceph.com/listinfo.cgi/ceph-users-ceph.com
_______________________________________________ ceph-users mailing list ceph-users@xxxxxxxxxxxxxx http://lists.ceph.com/listinfo.cgi/ceph-users-ceph.com
