Closed labilezhu closed 2 years ago
Hi @labilezhu. Thanks for your feedback.
This does look like a bug and I was able to reproduce it in our lab. I will include a fix for this in the next release.
At the same time, we generally do not recommend to use more than 256 luns per iSCSI target.
If you have high amount of volumes, I would highly recommend using automatic target-lun control provided by the driver, by setting dynamicTargetLunAllocation: true
in the driver config. You can also use numOfLunsPerTarget
to control the amount of luns associated with a target before creating new target.
All config options can be found here https://github.com/Nexenta/nexentastor-csi-driver-block/blob/master/README.md#configuration-options.
Thanks @Qeas . I will try dynamicTargetLunAllocation
Symptom
When ISCSI LUN(Logical unit number) > 255, MountVolume fail.
pod description:
nexentastor-block-csi-node
driver log:I can see two symbolic link naming pattens in
/dev/disk/by-path
on the host:We knows:
I can see that:
Root cause
May be a bug of
nexentastor-csi-driver-block
: https://github.com/Nexenta/nexentastor-csi-driver-block/blob/58d0124997759f47e82217d02fbf03e1198a4ed5/pkg/driver/nodeServer.go#L442Reference below issues:
The right logic should looks like:
https://github.com/systemd/systemd/blob/main/src/udev/udev-builtin-path_id.c#L58
Ref
https://bugs.launchpad.net/cinder/+bug/1493350 https://github.com/kubernetes/kubernetes/issues/48639 https://github.com/kubernetes/kubernetes/issues/45024 https://access.redhat.com/solutions/702413