conversations-backend CrashLoopBackOff Fix
Problem
The conversations-backend pod (conversations-backend-89fc95d94-dxldl) was in CrashLoopBackOff with exit code 137 (OOMKilled).
Root Cause
The pod was using the "micro" resource preset, which allocates only 384Mi memory limit (256Mi request). The application process exceeded this limit and was killed by the kernel OOM killer.
Resource preset values (defined in helmfile/apps/common/charts/common/templates/_resources.tpl):
| Preset | Memory Request | Memory Limit |
|---|---|---|
| nano | 128Mi | 192Mi |
| micro | 256Mi | 384Mi |
| small | 512Mi | 768Mi |
| medium | 1024Mi | 1536Mi |
Fix
Changed conversations.backend resource preset from "micro" to "small" in:
File: helmfile/environments/default/global.yaml.gotmpl:38
conversations:
- backend: "micro"
+ backend: "small"
frontend: "nano"
This increases memory from 384Mi → 768Mi limit and 256Mi → 512Mi request.
Redeploy
After applying the fix, redeploy the conversations app:
helmfile -e demo -l name=conversations apply