Best way to resolve bottlenecks with Flink?

classic Classic list List threaded Threaded
2 messages Options
Dan
Reply | Threaded
Open this post in threaded view
|

Best way to resolve bottlenecks with Flink?

Dan
My job has very slow throughput.  What are the best signals that will indicate if there are performance issues?

Is there an overall health summary that would indicate the most likely issues impacting performance?

I found a variety of pages and metrics.  I resolved some of the backpressure in my job.  I looked at memory issues.  My direct "Outside JVM" is full (but I'm not sure if that's normal).
Reply | Threaded
Open this post in threaded view
|

Re: Best way to resolve bottlenecks with Flink?

r_khachatryan
The closest thing is the backpressure status which you mentioned.
From there, you can troubleshoot specific subtasks by inspecting their metrics.
There is no health summary in Flink at the moment.

Regards,
Roman


On Fri, Sep 25, 2020 at 5:35 AM Dan Hill <[hidden email]> wrote:
My job has very slow throughput.  What are the best signals that will indicate if there are performance issues?

Is there an overall health summary that would indicate the most likely issues impacting performance?

I found a variety of pages and metrics.  I resolved some of the backpressure in my job.  I looked at memory issues.  My direct "Outside JVM" is full (but I'm not sure if that's normal).