How to trigger GKE autoscaling decision on demand?

ca flag

The GKE autoscaling process makes what they call "decisions" about whether or not to autoscale. The logs will include details explaining why the process decided not to do anything (noDecisionStatus), which is cool. Search bait: This can also be seen as a notification with the text "Can’t scale up a node pool because of a failing scheduling predicate".

I've made some changes that should address the issues raised in the noDecisionStatus log entry. How do I get GKE to try making a decision again? It's been hours.

vn flag
Same error. What changes have you made?
ca flag
Various, in response to what Google suggests. Things like changing pod tolerations. I still have volume node affinity conflicts (which I totally expected and aren't an issue, lol, because of course there are conflicts) but I haven't dealt with them yet as I'm waiting to see if the other changes have any effect.
ca flag

Looks like I just had to wait. I'm seeing the noDecisionStatus logs again.


Post an answer

Most people don’t grasp that asking a lot of questions unlocks learning and improves interpersonal bonding. In Alison’s studies, for example, though people could accurately recall how many questions had been asked in their conversations, they didn’t intuit the link between questions and liking. Across four studies, in which participants were engaged in conversations themselves or read transcripts of others’ conversations, people tended not to realize that question asking would influence—or had influenced—the level of amity between the conversationalists.