Provider doesn't exit when a kafka-connect cluster is on 100% CPU utilization

[dimisjim]

Provider does not exit when a kafka-connect cluster is unresponsive, leading to endless

[TRACE] dag/walk: vertex x is waiting for y

messages

This happens when I target multiple connectors of the same cluster, but not when I am targeting only 1 or a few of them.

[dimisjim]

There seems to be a threshold on the amount of connector resources where this happens.

If I target all of them, the provider endlessly keeps the plan going, but in a few occasions, it succeeds.

If I target 70% of them, it succeeds all the time.

Any clues on what might be the problem?

[pbr0ck3r]

I can second this. Have this issue when running v0.2.3, do you get this issue on v0.2.2? Also how many connectors are you running? We have 128 connectors we are trying to check via terraform plan.

[pbr0ck3r]

Seems to get to around 100 connectors then hang.

[dimisjim]

In my case it was around 70. But CPU was also 100%

[pbr0ck3r]

@Mongey do you have any thoughts on if this is a issue with code or possibly a issue/limitation from the terraform side (or machine CPU/Memory related)? If you can help point me in the right direction, I wouldn't mind spending some time on trying to dig into this and maybe create a PR if possible.