Werk #9165: Fixed processes accessing user volume blocking failover in some cases
| Component | Cluster |
| Title | Fixed processes accessing user volume blocking failover in some cases |
| Date | Oct 22, 2015 |
| Level | Prominent Change |
| Class | Bug Fix |
| Compatibility | Compatible - no manual interaction needed |
| Appliance Version | 1.2.2 |
In some cases it happened that processes started by users prevented failover of the cluster causing it to hang or fail.
For example when there are SSH clients (like WinSCP) which can instantly reconnect when they were killed may cause this issue when logged in as site user via SSH or as root and accessing the sites directories.
The cluster filesystem management code was already trying to end all processes accessing the cluster filesystem, but there was a race condition when the script send a kill/term signal to the processes and a new process appears accessing the cluster volume before it could be unmounted.