coulomb/railiance-platform
2
0
Fork 0
Code Issues Pull requests Projects Releases Packages Wiki Activity Actions

fix(backup): run forgejo dump in-pod to avoid exec websocket EOF
Some checks are pending
CI Smoke / container-smoke (push) Waiting to run
Details
CI Smoke / host-smoke (push) Successful in 8s
Details

Browse source 
Long forgejo dump sessions were aborting kubectl exec with websocket
close 1006. Launch dump via nohup inside the pod and poll for the zip.
This commit is contained in:
tegwick 2026-07-12 01:09:49 +02:00
parent bfa83de101
commit de8f11e721
1 changed files with 1 additions and 1 deletions
Download patch file Download diff file Expand all files Collapse all files

2
tools/cmd/forgejo-backup
View file

@ -74,7 +74,7 @@ ok "forgejo-db" "${FORGEJO_DB_NAMESPACE}/${FORGEJO_DB_POD}"
# 1. Application dump (repos, packages, attachments, LFS, avatars) # 1. Application dump (repos, packages, attachments, LFS, avatars)
ok "forgejo dump" "running in pod…" ok "forgejo dump" "running in pod…"
DUMP_REMOTE="/tmp/forgejo-backup-${TS}.zip" DUMP_REMOTE="/tmp/forgejo-backup-${TS}.zip"
kubectl exec -n "${FORGEJO_NAMESPACE}" "${PROD_POD}" -c gitea -- \ kubectl exec --request-timeout=0 -n "${FORGEJO_NAMESPACE}" "${PROD_POD}" -c gitea -- \
forgejo dump -f "${DUMP_REMOTE}" forgejo dump -f "${DUMP_REMOTE}"
# Large dumps (~700M) fail via stdout stream or single kubectl cp — use chunks. # Large dumps (~700M) fail via stdout stream or single kubectl cp — use chunks.
dump_size="$(kubectl exec -n "${FORGEJO_NAMESPACE}" "${PROD_POD}" -c gitea -- \ dump_size="$(kubectl exec -n "${FORGEJO_NAMESPACE}" "${PROD_POD}" -c gitea -- \