Check readiness
A cluster can have every service up and still be unable to do its job: no storage class, no object gateway, an operator image never built. These gaps are readiness findings. They are not faults, so each comes with its fix.
stratumctl admin readiness # the findings and their fixesstratumctl admin readiness --wait 30m # wait until ready; exit 1 if notConsole: Admin › Substrate › Overview. After an install, and for a while after a node joins or restarts, the cluster is settling: warnings are held back and the console shows “Starting: N of M”. Blocking findings are never held.
What a new cluster shows
Section titled “What a new cluster shows”| Finding | Severity | Means | Fix |
|---|---|---|---|
FormationNoFaultTolerance |
info | Fewer than three control-plane nodes: losing one stops writes. | Join the rest. |
FormationBelowPlannedSize |
info | Planned for five control-plane nodes, has three or four. | Join the rest. |
JoinCodeLiveAtPlannedSize |
info | A join code is live and no more nodes are expected. | Revoke it. |
StorageAwaitsCeph |
warning | No storage class, and Ceph is the compute storage. | Set up Ceph. |
NoStorageClass |
blocking | No storage class: instances and volumes cannot be created. | Create a pool and a class. |
ObjectStorageAwaitsCeph |
info | No Ceph, so no object gateway. | Set up Ceph, then the gateway. |
ObjectStorageNotDeployed |
warning | Ceph has no object gateway. | Deploy it. |
OperatorRecipeNeverBuilt |
warning | A managed service’s image was never built. | Build it. |
NoImages |
warning | Nothing to launch instances from. | Import an image. |
NoAlertChannel |
warning | Alarms have nowhere to go. | Set a channel. |
BackupsOff |
info | No platform backup. | Configure one. |
StoragePathMTU, OverlayPathMTU |
blocking | A network’s MTU is more than its path carries. | Lower the MTU or fix the path. |
NodeNotReporting |
warning | A node’s heartbeat lapsed. | Wake or repair it. |
MembershipLagging |
warning | A node has not applied the current membership 10 minutes after a change. | journalctl -u stratum-membership-converge on that node. |
Most clear by themselves once the gap closes. The rest of the list is under Troubleshooting and Health and readiness.
Next: set up storage.