Prasanna Kumar Kalever
Prasanna Kumar Kalever
@raghavendra-talur @jarrpa @phlogistonjohn please help review.
@Madhu-1 what is the OS version ? Could you give the output of, \# cat /etc/redhat-release; uname -a; rpm -qa | grep -e tcmu -e gluster-block -e targetcli -e rtslib...
@Madhu-1 the issue looks like something related to targetcli version or compatibility. I need to check the setup. Right now, I'm in between something. Will work with you later or...
@humblec, I think @Yuggupta27 is busy with CI/e2e stuff, I'm happy to take over. @Yuggupta27 feel free to jump in :-) Thanks!
Ack! Thanks for reporting Madhu, I have also encountered this before, but looks like it's not hit very frequently. I will investigate on the root cause and get back. PS:...
Sent a PR to rbd-nbd: https://github.com/ceph/ceph/pull/42609
> problem hit again [#2322 (comment)](https://github.com/ceph/ceph-csi/pull/2322#issuecomment-893155681) [ Copying Comment from the other thread, for context:] > > /test ci/centos/mini-e2e/k8s-1.21 > > `Aug 4 20:23:22.683: INFO: >>> kubeConfig: /root/.kube/config > Aug...
Hello @trociny Thanks for looking at this, much appreciate all the help. > @pkalever Actually I was pessimistic about your "sync" patch because it does not help much in general...
> > I.e. it is exactly 30 sec, and this is exactly the default rbd-nbd `--io-timeout`. > > @trociny What sets it? We don't have a default value for `--io-timeout`...
@idryomov There is a possible bug in this area, i.e. once we alter io-timeout (say for nbd0), the sysfs will keep it persistent and use the same for future mapped...