[SOLVED] SSH and web interface not working after reboot.

[DLZ-TVotR]

Hi,

I powered down my server to install a new dual NIC pci expansion card. I am not using that card yet, just installed it. Upon powering on the web interface and ssh fail. From looking at older posts I checked on the following:

pve-cluster.service is failed, and I cannot start it with it stating "Start request repeated too quickly":



/etc/hosts is correct




I can ping out to the internet and the servers IP is 192.168.0.55 (cut off the 1 in the screenshot, sorry). /etc/hostname shows pve and hostname -f shows pve.local

/var/log/daemon.log states it failed to load local private key (key_file or key) at /usr/share/per15/PVE/APIServer/AnyEvent.pm




Which makes sense since /etc/pve/local does not exist:



I mentioned SSH as well as that seems related since the authorized_keys file should be in /etc/pve/priv which also does not exist:



I have rebooted with no luck.

quick edit: no vms are running either

 

[DLZ-TVotR]

Looking at journalctl -u pve-cluster it appears the issue may be a mounting issue. I am not sure what I should be unmounting, and how to remount given its a fuse mount error:






Wondering if my issue is that drive I have mounted to /etc/pve/hdd ? It is listed in /etc/fstab but I don't recall adding it manually. I thought that was just added as local storage via the UI.

Removed from /etc/fstab, and removed the /etc/pve/hdd folder entirely (after making sure it was all that existed in /etc/pve) and rebooted


That worked like a charm


Admins/Mods: consider this solved. It is still pending approval before public posting but feel free to post, might help someone else. I still don't know exactly how I ended up here but that drive had nothing on it so I can re-create or do whatever as needed.

 

[Stoiko Ivanov]

That worked like a charm

Glad you found the issue!

the mount in /etc/pve is quite odd though (nothing in PVE's tooling would create that) - maybe some manual creation at some time before the last reboot?

In any case - having a mountpoint inside /etc/pve is nothing that will work smoothly (unless you schedule that mount to be done after the pmxcfs starts, but I would suggest to simply use some other point of the directory-tree as mountpoint (e.g. /mnt/pve/) - since this causes far less confusion)

I'll mark the thread as 'SOLVED' (you can do that as well by clicking on '...' 3 dots above your first post -> Edit Thread -> and selcting 'SOLVED' as prefix - just as info for the next time)

Thanks for sharing your solution!