weizhouapache commented on PR #14320: URL: https://github.com/apache/cloudstack/pull/14320#issuecomment-6217862821
Manual test: force-stopping kvm2, with the heartbeat objects named after the GUID of the host (code as in 78e4a2f8f7d). **Environment:** nested test environment with 2 management servers, 2 KVM hosts (Ubuntu 26.04) and RBD-only primary storage on Ceph (2 pools), `force.ha=true`, 4 user VMs on kvm2. `test_hostha_kvm.py` passed 9 of 9 just before, on the same build. **Before the test:** both hosts `Up`, and the pool holds a fresh `KVMHA-hb-<uuid>-LibvirtComputingResource` object for each host. **Test:** the VM which backs kvm2 was force-stopped in the parent lab (`stop virtualmachine forced=true`). Times are CEST (2026-10-16); the agent logs are in UTC, 2 hours earlier. | Time (CEST) | Event | After the stop | |---|---|---| | 14:35:37 | kvm2 VM force-stopped in the parent lab | 0 s | | 14:39:05 | kvm2 `Down` in the test environment | 208 s | | 14:39:10 | the 4 VMs go to `Starting` on kvm1 | 213 s | | 14:41:23 | all 4 VMs `Running` on kvm1 | 346 s | | 14:42:03 | kvm2 VM started again in the parent lab | | | 14:43:46 | kvm2 `Up` again | | **Checks:** in the agent log of kvm1 the heartbeat check of kvm2 ran as `kvmheartbeat_rbd.py -s <monitor> -o <pool> -n <user> -h 107fcb59-6915-3f29-9cbc-cb8d4972eb20-LibvirtComputingResource` (the GUID of kvm2, as the management server has it) and reported the host as dead; no Ceph key is in the arguments. The host was only reported `Down` after those checks, and the VMs were restarted by HA on the surviving host. -- This is an automated message from the Apache Git Service. To respond to the message, please log on to GitHub and use the URL above to go to the specific comment. To unsubscribe, e-mail: [email protected] For queries about this service, please contact Infrastructure at: [email protected]
