Cloud Run Instance Quota at 100%

[thameezbo]

Hi 

When running a cloudrun job with 86 executions (tasks set to 1, parallelism not configured, resources with cpu = "4000m" and memory = "8Gi" ) we hit the Instance limit per region quota.

This limit then stays for hours, even after all services and jobs have been deleted. 

Europe-west3

[knet]

Hi, did this issue start recently? I think it's related to the issues around jobs taking a long time to start, but I'll raise this to the engineering team.

[thameezbo]

We've only recently started working with jobs so not really certain
unfortunately.

However two things that would be great to figure out:
- the relation between resources (CPU and memory) and instance count (it
seems that higher resources equals to a higher instance count, but what is
the actual ratio at play here?)
- why the cool-down period for the quota remains high for hours even after
all cloudrun assets are deleted (both services and jobs)

[knet]

An instance is 2vCPU / 2GB of memory. 

Regarding the cooldown, I wasn't aware of that. We'll look into it. 

[knet]

Correction: an instance is 1vCPU / 2GB of memory.

We are working on getting rid of the instance quota; in the future, the CPU and Memory quotas will be the ones that matter.