search

shapeblue / Trillian

Trillian - flexible monkey powered CI/CD
Apache License 2.0
14 stars 13 forks source link

During Trillian env deployment ensure SSVM is Up #33

Closed rohityadavcloud closed 1 year ago

rohityadavcloud commented 7 years ago

Several env deployment failures occur (esp in case of KVM) when SSVM agent is not in UP state and fails to setup the default built-in template. Without this, the zone deployment fails. We can have a workaround to poll and ensure that SSVM agent state is UP (after the VM state is Running) and have a timeout/retries where in case of failure we try to destroy the SSVM and perhaps a new SSVM can come up cleanly.

/cc @PaulAngus @DagSonsteboSB

borisstoyanov commented 7 years ago

Thanks @rhtyd, I'll try to add this check

borisstoyanov commented 7 years ago

@rhtyd can you give an example of the failures you're getting. I see the current steps are waiting the system vms to come 'Running' and then wait the agent status to be 'Up' https://github.com/shapeblue/Trillian/blob/master/Ansible/roles/cloudstack-manager/tasks/wait-for-environment.yml

rohityadavcloud commented 7 years ago

This is one of those cases where agent is not in Up state. I've hit this couple of times, let me ping a case the next time it happens @Bobby

borisstoyanov commented 7 years ago

but, if you see the check in the link, it actually does a 200 retries if the agent is Up on each system vm. I think this check is already there.. Or.. maybe the agent is Up and goes away, which fails the template download, I think I've seen this as well..