Flink restarts after hitting outOfMemory

[Misha Hungyo <mk...@gmail.com>]

Hi,
I have a flink job that runs continuously(the job never goes to* FINISHED*
state but is always in *RUNNING* state) and there are stateful operations
too. But the states are cleared after the time threshold is hit. Logically,
from what I understand, the memory claimed should be cleaned up but this
doesn't seem to happen. After few hours, the taskmanager throws
the OutOfMemoryError and jobmanager restarts.

How can I improve or solve this issue?

-- 
Misha Hungyo
R&D Nokia Networks

Re: Flink restarts after hitting outOfMemory

[Caizhi Weng <ts...@gmail.com>]

Hi!

Heap memory usage depends heavily on your job and your state backend. Which
state backend are you using and if possible could you share your user code
or explain what operations your job is doing?

Misha Hungyo <mk...@gmail.com> 于2022年2月14日周一 16:56写道：

> Hi,
> I have a flink job that runs continuously(the job never goes to* FINISHED*
> state but is always in *RUNNING* state) and there are stateful operations
> too. But the states are cleared after the time threshold is hit. Logically,
> from what I understand, the memory claimed should be cleaned up but this
> doesn't seem to happen. After few hours, the taskmanager throws
> the OutOfMemoryError and jobmanager restarts.
>
> How can I improve or solve this issue?
>
> --
> Misha Hungyo
> R&D Nokia Networks
>