- Shipped
- August 27, 2026 at 5:33 PM UTC
- Author
- Kamo
- Commit
- 5962489
FreeSWITCH runs with -nc, so mod_logfile is the only record of what it did, and on the container filesystem that record dies with the container — precisely when it is needed. A segfaulting FreeSWITCH left nothing at all to read. The log now lives on a pod-scoped emptyDir, mounted read-only into the ecallmgr container as well: `kubectl exec` into a crash-looping container races the restart and usually loses, while ecallmgr stays up in the same pod. This is what finally exposed the crash — a null event profile from a malformed upstream default.xml, fixed in kamo-pbx.