GKE sometimes dont stop instance-groups

[julian21olarte]

Hi, sometimes Zorya dont stop the clusters (instance-groups), I have added some prints to check what is happening and I could see that the function in gcp/gke.py:
no_of_nodes = gcp.get_instancegroup_no_of_nodes_from_url( url )
usually return the instance-group size zero, then Zorya not work because it cant resize to zero a instance-group that already has zero nodes.... to fix that temporally I need to run the Zorya cron job manually many times to get all clusters down, thanks for any help.

[eranchetz]

@julian21olarte thanks for finding this, let me try and reproduce this on my local environment and see if I get the same results.

[hungpv29]

I have some questions:

• If Minimum size in the node-pool is >0 the resize action will probably fail
• If autocscaling is on in the node-pool, I do not know what will happen with the reduction size command

[okiirawan]

@eranchetz Is there any workaround for this issue?

[eranchetz]

@okiirawan we are working on that, hopefully we will have an update soon