On 10/18/19 4:44 AM, Martin G wrote:
I ran into trouble connecting VMWare 6.7 to CEPH cluster exported via RGW and Ganesha.
Once the datastore is created, the ganesha thread is stuck at 100% CPU in
mdcache_find_keyed_reason, preceeded by pthread mutex lock and unlock.
I am at Nautilus, deployed by Ansible, running on latest CentOS 7. The issue appeared in
Ganesha 2.7.3 and also in 2.8.
Stacktrace from perf top here: https://imgur.com/a/rQ3AKfy
I would greatly appreciate any clues or hints on what causes this.
I think you want
92b4a45920da15a6f9d8578de9dd552023408caa MDCACHE - Try to recover chunk
after dropping lock
c1c70e813f87e22c4cd7fa355aecf82d5a61a818 MDCACHE: save all dirents from
last READDIR in dir-map
Can you try with 3.0, which has these?