Problem deploying multiple Flink streamlets in one go about cloudflow HOT 5 CLOSED

Thanks @mrooding - I will take a detailed look at the logs and get back.

from cloudflow.

Tried to replicate the scenario by adding more flink streamlets to taxi-ride application. Could not replicate - deploys fine.

@mrooding - can we have the operator log files when this problem occurred ?

from cloudflow.

Hi @debasishg

Please find the logs attached. Not sure if you saw my PM on Gitter but I'll just leave that here too:

I'm starting to think that it may have something to do with the flink operator which keeps erroring on the "deleting" phase

Somehow, I only see this output for the flink streamlet which fails, the other flink streamlet consistently works fine:

{"json":{"app_name":"stock-market-app-rfq-prediction-calculator","ns":"stock-market-app","phase":"Deleting"},"level":"warning","msg":"Failed to reconcile resource stock-market-app/stock-market-app-rfq-prediction-calculator: GetJobOverview call failed with status FAILED and message '': Get http://stock-market-app-rfq-prediction-calculator-014a2e45.stock-market-app:8081/jobs/88606551181e331e44ffb1d3c2268f66: dial tcp: lookup stock-market-app-rfq-prediction-calculator-014a2e45.stock-market-app on 10.19.240.10:53: no such host","ts":"2020-03-02T15:44:57Z"}

flink-operator.log

cloudflow-operator.log

from cloudflow.

@mrooding I have been trying to reproduce this bug on Cloudflow 1.3.1 and I don't observe any related issues. All Flink streamlets seem to deploy correctly.

Are you still observing this issue? If so, could you use v1.3.1 and check? We have upgraded to Flink 1.10 and Flink Operator 0.8.2.

Maybe this issue is solved as a side-effect of that?

from cloudflow.

This seems to work on 1.3.1 so I will close this issue.

from cloudflow.