Hello,

we had this issue previously - it was connected to timeouts, where the socket disappeared due to a timeout before a reply could be sent back. In our case this was caused by having link-local multicast name resolution (LLMNR) on by default in systemd-resolved, which was evident by slow calls to `getent hosts <hostname>`.

Hope this helps,
Tilman Hoffbauer

Hello everyone,

 

Not sure if you guys have heard this tune already but did anyone come across a solution for “Unexpected missing socket error”.
There is nothing useful in the logs but the message appears on compute nodes and slurm controller node.

 

Thank you,

 

Vlad Ozeryan

AMDS – AB1 Linux-Support

Vladimir.Ozeryan@jhuapl.edu

Ext. 23966