Failed deactivating swap /dev/pve/swap

[silvered.dragon]

Hi to all,
in a 3 node ceph cluster buit on 3 identical hp gen8 dl360p servers, I'm always receiving the error attached everytime I'm rebooting a node. Before rebooting the node I always move all the vms to another node, so when I press reboot there is no running VM. To fix this I have to force an hardware restart, cause after 30 minutes it still stucks on this error.
Is there a way to fix this? many thanks
**edit: the error happens with all those HPs servers but in the same cluster I have another node without ceph, only for utility stuffs, and this will reboot without problems.

 

[Chris]

Hi,
please try to check the syslog for errors. Also you could try to run a systemctl stop swap.target before you reboot the system and check with systemctl status swap.target for errors.

 

[silvered.dragon]

Hi,
please try to check the syslog for errors. Also you could try to run a systemctl stop swap.target before you reboot the system and check with systemctl status swap.target for errors.

Thank you @Chris for the reply,
attached you can find the syslog in a timestamp around the reboot process,

if you check at

Feb 15 17:58:20 nodo1 systemd[1]: Stopping PVE API Daemon...

this is exacly when it stucks, but unfortunately I cannot see any errors.

For errors checking with systemctl, I have to wait the week cause this is a production environment and I really don't want to test things during the job even if I can migrate VMs to other nodes, I will update this thread saturday.

thank you

 

[thheo]

Hi,

I face the same thing on an Intel server

 

[JuniPeruvian]

Hi,
I have the same problem.
any solution about it?

 

[silvered.dragon]

I noticed that if I disable write cache on my raid controller the issue seems to disappear

 

[JuniPeruvian]

I noticed that if I disable write cache on my raid controller the issue seems to disappear

I get it,
and what would be the command to perform the operation that worked for you?

 

[silvered.dragon]

I get it,
and what would be the command to perform the operation that worked for you?

for me there is a web interface that manages the raid configuration, but I don't want to keep this setting disabled so this doesn't solves the issue for me

 

[JuniPeruvian]

for me there is a web interface that manages the raid configuration, but I don't want to keep this setting disabled so this doesn't solves the issue for me

I get it.
Thanks for answering.

 

[papete]

Here, same problem using 2 nodes with cephs. I has been tryed to do that Chris say then reboot fine. Antother try witout it and fine too.

I'll be watching.

 

[silvered.dragon]

Here, same problem using 2 nodes with cephs. I has been tryed to do that Chris say then reboot fine. Antother try witout it and fine too.

I'll be watching.

lately I've noticed that if I run these commands AFTER SHUTTING DOWN ALL THE VMS so you will have enough free phisical ram to make things works

swapoff -a

wait some seconds

swapon -a

and then reboot from web interface, will not hung during the reboot process.. at least in this way I can reboot the nodes from remote without hard reset the servers

 

[papete]

Tip noted. thank you.

If it is a bug maybe they should be reviewed for next release.

Can someone notify it? I'm a novice and I don't know how.

 

[sahostking]

busy rebooting one server now and noticed this same issue using Dell R710 server.

 

[izedg]

Intel Scalable

 

[sahostking]

yes its still happening it seems. I just do swapoff before I reboot now but that takes ages soemtiems.

 

[elmarconi]

After kernel upgrade I did a Hibernate on the VM's and a following reboot hung with the same msg as the above shown by izegd.
Had to powercycle the host.

 

[EmilioMoreno]

Same error here, latest pve6 version. After 1 hour I reboot the server

 

[hk@]

reporting in - here seeing the same thing (latest pve6 - yes it's unsupported now....)

 

[brucexx]

Similar message here , 5 node cluster with Dell Servers on Intel CPUs, fully updated version 6 with community subscription. I had to reboot two servers for maintenance in last two month and each had this message:

Failed deactivating swap /dev/pve/swap
A Stop job is running for /dev/dm-8 (8 min..../ no limit)

Once I saw no limit and already 8 min wait time I hard rebooted from iDrac.