Handle datastores with spaces or special characters #1116
Comments
@ashahi1 can you check that the host is having the datastores on it. The exception is not handled in vmdk_utils.py:get_datastore_name() and should be fixed, but the ESX service is complaining that local-0 and local-1 datastores aren't available, 03/30/17 07:43:07 463428 [MainThread] [WARNING] dockvols creation failed - No such file or directory on datastore local-0 (11) Can you check the ESX host to see the datastores are present and accessible. Run esxcfg-scsidevs -m |
And for all these tests its a good thing to run the ESX service with log level set to DEBUG in /etc/vmware/vmdkops/log_config.json. The trail is easier to debug with values of variables logged. |
@govint I checked from the VC and all the host's datastores look fine. I just ran a test that created 800 Volumes on the same setup but didn't see any issue.
|
@ashahi1 can you also get the docker host system logs (journalctl --no-pager --system > ). The plugin seems to have restarted when removing volumes. Was that a manual restart? 2017-03-30 05:59:19.292149597 +0000 UTC [INFO] Removing volume name="dockerVol_fONzs" May not be related to this issue,. just want to know if there was a crash and restart of the plugin. |
@govint No, I did not do any manual restart of the plugin. Host system logs: |
No plugin panic in the logs one explanation for this behavior is like this:
|
From the exception stack, it looks like the VM datastore is "for some reason" not in the datastores list - in mem. cache - used by the service. It may be local-0 or local-1 because along with the exception stack we also see the messages related to dockvols path missing on local-0 and local-1. 03/30/17 07:43:04 463408 [MainThread] [WARNING] dockvols creation failed - No such file or directory on datastore local-0 (11) 03/30/17 07:43:06 227484 [Photon-Dker-8-nolock-list] [WARNING] dockvols creation failed - No such file or directory on datastore local-1 (11) Why the datastore isn't in the datastores list is because if the dockvols can't be created (which is called from init_data_stores in vmdk_utils.py) then we ignore the datastore. Any request from a VM on that datastore is always going to fail and with an exception. The dockvols folder isn't getting created because the datastore is named as "local-0 (11)", which causes the mkdir command to fail, as reproduced locally below, with a datastore named "bigone (1)". 04/03/17 09:58:33 1169339 [Thread-4] [WARNING] dockvols creation failed - No such file or directory on datastore bigone (1) |
Customer is running into this issue. Bumping the priority. |
Fixed via #1140 |
ESX has 10 VMs and I am trying to create volumes from each VM simultaneously.
Volume creation failed from one of the VMs:
docker volume ls:
Logs:
vmdk_ops.1.txt
vmdk_ops.txt
docker-volume-vsphere.txt
The text was updated successfully, but these errors were encountered: