This message was deleted.
# atlantis-community
s
This message was deleted.
1
j
This is likely not atlantis but terraform consuming the resources. Atlantis is just a runner and webhook responder right? You can adjust the number of concurrent operations that terraform is allowed to perform (the default seems to be 10) https://developer.hashicorp.com/terraform/cli/commands/plan#parallelism-n
Setting this value to less than 10 will result in longer running of plan/apply but may help with how "spiky" the resource utilization is.
r
isn't the parallel pool size on a per-PR basis and not global? if so, perhaps a feature to set upper limits on concurrent global processes could also be interesting. i may be misremembering the code.
m
Thanks @Jason Reslock for your answer. You're right terraform does consume the resources, not Atlantis. I will study this option but my idea was more to get a feature to spread the resource usage over several pods (so on several nodes).
j
@Maxime Valo horizontal scaling would be ideal. There is an open discussion here that has not seen much activity but it is a good starting point for talking about what it would take to implement such a feature.
p
there is a lot of issues related to scalability and HA
the other problem is that we (the maintainers ) can't implement it without the help of the community and coordinating the process would have tricky
and it needs delelopers commuted to maintain it , etc
m
ok I see this is definitely not an easy-to-do improvement.