Page MenuHomePhabricator

labs-vagrant broken due to NFS issues
Closed, ResolvedPublic


Two labs-vagrant boxes died today at the same time with NFS issues:

vagrant@mediawiki-vagrant:~$ ls /vagrant 
ls: cannot access '/vagrant': Protocol not supported
tgr@wikispore-test:/srv/mediawiki-vagrant$ vagrant reload
The following SSH command responded with a non-zero exit status.
Vagrant assumes that this means the command failed!

mount -o vers=3,udp,noatime,rsize=32767,wsize=32767,async /vagrant

Stdout from the command:

Stderr from the command:

mount.nfs: mount to NFS server '' failed: RPC Error: Unable to receive


vagrant config nfs_force_v4 yes
vagrant provision
vagrant reload

Event Timeline

Tgr updated the task description. (Show Details)

vagrant config nfs_force_v4 yes seems to fix it, the error goes away and /vagrant seems to work correctly inside the guest VM. The wiki is still broken though.

vagrant config nfs_force_v4 yes
vagrant provision

together seems to have helped.

Krenair removed a subscriber: Krenair.
Krenair added a subscriber: Krenair.

Sounds like fallout from the changes I made for T257855: Problems with NFS shares when launching a mediawiki-vagrant instance in Cloud VPS. @Tgr did this happen after you updated the git clone in /srv/mediawiki-vagrant for an existing deployment of ::role::labs::mediawiki_vagrant?

No, it happened without any manual changes to the boxes (which were indeed existing deployments of ::role::labs::mediawiki_vagrant). I remember I had to make a git pull for the nfs_force_v4 vagrant option to be available (ie. the vagrant checkout was older than rMWVA48725d50e989: Allow forcing NFSv4 between host and vm and not auto-updating; it was probably last updated in late spring).

Interesting. I wonder if a new kernel is actually the underlying cause of the NFSv3 mounts starting to fail? The fix should be just what you did, so I'm thinking at most this is something to document as a troubleshooting step on the wikitech page.

Tgr claimed this task.