Closed dcz010 closed 2 months ago
Show your systemd unit and socket file.
I had a problem when I was using dynamic users with systemd. Systemd had a bug where it wasnt cleaning temporary folders and files and I ran out of inodes.
Hi, Thanks for your reply. Well i found the solution after long searching myself. The problem was, that the sockets weren't closed after and successful or failed connection and then it had an buffer overflow. I had to change the clusterchk@.service like that:
[Unit]
Description=Check the status of Galera/MySQL
After=mysql.service
Requires=clusterchk.socket
[Service]
Type=simple
ExecStart=-/usr/local/bin/clusterchk.sh
TimeoutStopSec=5
StandardInput=socket
#StandardError=journal
#Restart=on-failure
[Install]
WantedBy=multi-user.target
Hi,
Could there be an problem with the clusterchk.socket and systemd by filling up the system with so many file descriptors that the system can't no longer handle it?
Here are some syslogs from a debian vm:
It leads to an very slow system without the possibility to reboot (only hard reset at vm level possible).
Greetings dcz010