General

Profile

Xiubo Li's activity

From 06/30/2021 to 07/29/2021

07/29/2021

09:23 AM CephFS Bug #51956 (Fix Under Review): mds: switch to use ceph_assert() instead of assert()
Xiubo Li
09:21 AM CephFS Bug #51956 (Resolved): mds: switch to use ceph_assert() instead of assert()
If the -DNDEBUG was specified when build the code, the assert() will do nothing.

I hit one odd issue that when...
Xiubo Li

07/28/2021

05:56 AM Ceph Revision 361ee535 (ceph): qa: multifs already enabled as default
Since pacific already mark multifs enabled as defaut.
Signed-off-by: Xiubo Li <xiubli@redhat.com>
Xiubo Li

07/27/2021

01:54 AM Ceph Revision a448d1c3 (ceph): qa: wait pgs to be clean before using the pools
Or in some use cases, like the mds-full tests, we will hit the
"PG_AVAILABILITY" warning.
Signed-off-by: Xiubo Li <x...
Xiubo Li
01:48 AM Ceph Revision f4288f2a (ceph): qa: move the is_full() to parent class
Signed-off-by: Xiubo Li <xiubli@redhat.com> Xiubo Li
01:48 AM Ceph Revision c7837484 (ceph): qa: set the object_size to 1M
Set the object_size to 1MB to make the objects destributed more even
among the OSDs.
Fixes: https://tracker.ceph.com...
Xiubo Li
01:48 AM Ceph Revision c1cea712 (ceph): qa: use the pg autoscale mode to calcuate the pg_num
Setting the pg_num to 8 is too small that some osds maybe not covered by the
pools, some osds maybe overloaded. Remov...
Xiubo Li
01:48 AM Ceph Revision a96ee419 (ceph): qa: do not raise for kclient for _fsync test
For kclient, the write() will return -ENOSPC instead of the fsync().
Fixes: https://tracker.ceph.com/issues/45434
Si...
Xiubo Li
01:48 AM Ceph Revision ba3833a6 (ceph): qa: do not multiple the full ratio twice
The cluster has already multiple the full ratio before returning
the "max_avail".
Fixes: https://tracker.ceph.com/is...
Xiubo Li
01:48 AM Ceph Revision 999c787a (ceph): qa: wait more time since there have many more pgs than before
Signed-off-by: Xiubo Li <xiubli@redhat.com> Xiubo Li
01:48 AM Ceph Revision 3456ff26 (ceph): qa: ignore PG_RECOVERY_FULL and PG_DEGRADED for mds-full
Signed-off-by: Xiubo Li <xiubli@redhat.com> Xiubo Li

07/26/2021

06:29 AM CephFS Backport #51285 (Resolved): pacific: mds: unknown metric type is always -1
Xiubo Li

07/20/2021

03:15 AM Linux kernel client Bug #51734: kcient: the delayed_work() send too many client_caps requests one time and will delay...
6740078 2021-07-20 11:05:41.660 149df03bb700 20 -- v1:10.72.47.117:6819/926278509 done calling dispatch on 0x558e1ef... Xiubo Li
01:20 AM Linux kernel client Bug #51734: kcient: the delayed_work() send too many client_caps requests one time and will delay...
This is from https://bugzilla.redhat.com/show_bug.cgi?id=1974882 Xiubo Li
01:19 AM Linux kernel client Bug #51734 (Won't Fix): kcient: the delayed_work() send too many client_caps requests one time an...
I have enable the kernel debug logs and hit one case that, the rmdir request was issued immediatelly just after the l... Xiubo Li

07/19/2021

05:49 AM CephFS Bug #51722 (Fix Under Review): mds: slow performance on parallel rm operations for multiple kclients
This is from bugzilla https://bugzilla.redhat.com/show_bug.cgi?id=1974882.
Xiubo Li
05:46 AM CephFS Bug #51722 (Resolved): mds: slow performance on parallel rm operations for multiple kclients
There has another case that could cause the unlinkat to be delayed for a long time sometimes, such as for the "remova... Xiubo Li

07/05/2021

02:23 AM CephFS Feature #51518 (Fix Under Review): client: flush the mdlog in unsafe requests' relevant and auth ...
Xiubo Li
01:35 AM CephFS Feature #51518 (Resolved): client: flush the mdlog in unsafe requests' relevant and auth MDSes only
Do not flush the mdlog in all the MDSes, which may make no sense for specific inode. Xiubo Li
01:55 AM Ceph Revision 0fc091f3 (ceph): client: flush the mdlog in unsafe requests' relevant and auth MDSes only
Do not flush the mdlog in all the MDSes, which may make no sense for
specific inode.
Fixes: https://tracker.ceph.com...
Xiubo Li
 

Also available in: Atom