Adding RBD disk with ValueError: No JSON object could be decoded about ceph-iscsi-cli HOT 6 CLOSED

with the failure in rtslib, you should check the versions of the kernel, python-rtslib and tcmu-runner are all correct.

from ceph-iscsi-cli.

Try using pdb to debug step by step to find out the root cause :)

from ceph-iscsi-cli.

When I use targetcli to create iscsi target, everything is OK. However, I get the above error when using gwcli.

And I get following error in /var/log/tcmu-runner.log

2017-11-21 09:58:09.601 894 [WARN] timer_check_and_set_def:229 rbd/iscsi-pool.ha-image: osd op timeout ((null)) must be larger than osd heartbeat grace (20) + interval (6)!
2017-11-21 09:58:09.601 894 [WARN] timer_check_and_set_def:248 rbd/iscsi-pool.ha-image: Will set the osd op timeout to 31 instead!
2017-11-21 09:58:09.641 894 [ERROR] tcmu_rbd_image_open:298 rbd/iscsi-pool.ha-image: Could not open image ha-image/osd_op_timeout=30. (Err -2)
2017-11-21 09:58:09.645 894 [ERROR] add_device:496: handler open failed for uio0

@pcuzner

from ceph-iscsi-cli.

The image name should be ha-image. However, there is a suffix /osd_op_timeout=30.

What is it? How could I solve the problem? Thanks!

from ceph-iscsi-cli.

the osd_op_timeout is picked out by tcmu-runner. Again I'm wondering if this is a version mismatch issue. the tcmu-runner code has a change which updated the format of the config string passed from rtslib - and if that doesn't match properly it will be a malformed request to tcmu-runner.

from ceph-iscsi-cli.

Thanks a lot! @pcuzner

from ceph-iscsi-cli.