Conversation
There was a problem hiding this comment.
I see the logic was: Run multipathd, if it fails (returns with error) - exit immediately with the error. Otherwise: check if the output contains contains the volumeIdVariation. If it does - exit. If not - sleep for intervalSeconds and retry - for maxRetries.
The logic after your change is: run multipathd. If it fails - sleep and retry. Otherwise - same as above, return if found the volumeIdVariation or else retry.
That's a change of the logic. Why?... Who says we want to retry in case of multipathd error?
| isNvme, err := d.NodeUtils.IsNvmeBaseDevice(baseDevice) | ||
| if err != nil { | ||
| return nil, status.Errorf(codes.Internal, "Multipath -f command failed with error: %v", err) | ||
| logger.Warningf("Failed to check if device %s is NVMe, assuming non-NVMe: %v", baseDevice, err) |
There was a problem hiding this comment.
this is... a problem. But I'm not sure there's a better way.
What would happen here? Either we're not NVMe, and then it's ok, or we may be NVMe and still try multipath -f, and it will fail, and so we won't clean the device.
What does actually clean the device eventually for NVMe? The function ClearStageInfoFile? Or what?...
shlomitn
force-pushed
the
release-1.13.1
branch
5 times, most recently
from
8a5fa1f to
222840f
Compare
shlomitn
force-pushed
the
release-1.13.1
branch
5 times, most recently
from
aa1a5a2 to
c59c90a
Compare
3 participants
In case of nvme, we don't need to run "multipath -f", we just need to escape it.
Once volume is unstaged, nvme devices are themselves removed from the multipath devices list.
For checking whether a device is nvme or not, I am checking existence of /sys/block/devicename/device/subsysnqn
(
nvme disconnectwould have removed all multipath nvme devices, so that was not exercised).