Dear Nebula Community Team
I have a problem with migrate vm after host error. I’m preparing test where I turn off server node01.nebula.infra (kvm node). Fencing and hook seems run correctly.
[2022-03-31 14:11:49 +0200][HOST 20][I] Fencing enabled
[2022-03-31 14:11:49 +0200][HOST 20][I] Success: Already OFF
[2022-03-31 14:11:49 +0200][HOST 20][I] Fencing success
[2022-03-31 14:11:49 +0200][HOST 20][I] states: 3, 5, 8
[2022-03-31 14:11:49 +0200][HOST 20][I] vms: ["328", "327"]
[2022-03-31 14:11:49 +0200][HOST 20][I] resched 328
[2022-03-31 14:11:50 +0200][HOST 20][I] resched 327
[2022-03-31 14:11:50 +0200][HOST 20][I] Hook finished
Host move to error state and vm start migrating. But i got below error:
Thu Mar 31 14:15:11 2022: Error executing image transfer script: VM named not found. /var/lib/one/remotes/datastore/xpath.rb:70:in `block in <main>': undefined method `elements' for nil:NilClass (NoMethodError) from /var/lib/one/remotes/datastore/xpath.rb:64:in `each' from /var/lib/one/remotes/datastore/xpath.rb:64:in `<main>' /var/lib/one/remotes/tm/ceph/mv: line 60: [: : integer expression expected ERROR: mv: Command "Get .monitor" failed: ssh: connect to host node01.nebula.infra port 22: No route to host INFO: mv: Moving node01.nebula.infra:/var/lib/one//datastores/124/328 to node03.nebula.infra:/var/lib/one//datastores/124/328 ERROR: mv: Command "set -e -o pipefail tar -C /var/lib/one//datastores/124 --sparse -cf - 328 | ssh node03.nebula.infra 'tar -C /var/lib/one//datastores/124 --sparse -xf -' rm -rf /var/lib/one//datastores/124/328" failed: ssh: connect to host node01.nebula.infra port 22: No route to host Error copying disk directory to target host
ssh from frontend to node03 is correctly. All fronted are ceph access.
I use Ceph datastore but It’s seems that OpenNebula want to ssh to host what is down. At my opinion It’s make no sens. I think it’s not correctly behavior. Why is that?