-
Notifications
You must be signed in to change notification settings - Fork 95
if datastore is renamed, all dockvols/<vmgroup> links are broken #1659
Comments
likely to be regression ... Please refer #818 where rename datastore related workflow was fixed and tested as well. |
This is what I'm seeing,
debugvol19 sharedVmfs-02 _DEFAULT 100MB 15MB ext4 N/A thin worker1-VM1.0 read-write independent_persistent worker1-VM1.0 Wed Jul 26 11:04:21 2017
debugvol19 sharedVmfs-3 _DEFAULT 100MB 15MB ext4 N/A thin worker1-VM1.0 read-write independent_persistent worker1-VM1.0 Wed Jul 26 11:04:21 2017
11111111-1111-1111-1111-111111111111 _DEFAULT This is a default vmgroup _VM_DS
debugvol19 sharedVmfs-3 _DEFAULT
debugvol19 sharedVmfs-0 _DEFAULT
root@photon-machine [ /vol/vpl-upg/docker-volume-vsphere/client_plugin ]# docker run -it --rm -v tvol11:/mnt busybox
This is working as expected and hence closing. |
When ESX joins VC cluster (and I suspect on other occasions) a datastore can get renamed.
In my case, datastore1 got renamed to datastore7, and then _DEFAULT link in dockvols got broken since it point to non-existing DS:
the end result is somewhat unpredicatable behavior - many ops will succeed since they use vmgroup UUID, and some may fail with messages about non-exiting files or folders
We should use ./ instead of /fullpath/uuid when forming links
@lipingxue @tusharnt
The text was updated successfully, but these errors were encountered: