ceph-system: run radosgw as a deployment #583
Reference in New Issue
Block a user
Delete Branch "benvin/ceph-system-rgw"
Deleting a branch is permanent. Although the deleted branch may continue to exist for a short time before it actually gets removed, it CANNOT be undone in most cases. Continue?
Part of #582. RGW runs on VMs (ausyd1nxvm2115-2119) and needs to move into k8s under GitOps; this adds the
ceph-systemnamespace with an in-cluster RGW behind a ClusterIP service. Cutover comes later.ceph-systemapp to the platform applicationsetclient.rgw.syd1section and explicit keyring pathkubernetes/namespace/ceph-system/default/rgw-keyring(keyring)quay.io/ceph/ceph:v18.2.8radosgw deployment, non-root, read-only root, startup probe, hard per-node spread, PDB + VPArgwon 7480Requires unkin/terraform-artifactapi#78 merged first.
ScheduleAnywayspread); all 3 RGW pods can land on one node → useDoNotScheduleon the hostname spread (keep a zone constraint soft if zones are uneven).--setuser ceph --setgroup cephare redundant underrunAsUser: 167non-root → drop them.vpa.yaml; cephrgw-system and 12 other base apps ship one → add for consistency.rgw-keyring(see cephrgw-system sibling); pods sit in ContainerCreating until seeded → add a one-line seed note.[client.rgw.syd1]sets nokeyring; the keyring is mounted at/var/lib/ceph/radosgw/ceph-rgw.syd1/keyringand found only via implicit$rgw_datadefault, otherwise radosgw looks in/etc/ceph/ceph.client.rgw.syd1.keyringand fails auth → addkeyring = /var/lib/ceph/radosgw/ceph-rgw.syd1/keyring(and optionallyrgw_data) to the section./(failureThreshold ~30, period 5s) and drop liveness initialDelaySeconds.quay.io/ceph/cephpull depends on the unmerged terraform-artifactapi allowlist; merging first yields ImagePullBackOff on 3 pods → link the tfartifactapi PR in the body and merge that first.No findings.