[ 
https://issues.apache.org/jira/browse/IGNITE-20449?page=com.atlassian.jira.plugin.system.issuetabpanels:all-tabpanel
 ]

Vyacheslav Koptilin updated IGNITE-20449:
-----------------------------------------
    Description: 
The failure handler should be able to create and log a thread dump on failure.

The failure processor configuration should be extended with the following 
fields:

 - `dumpThreadsOnFailure` This property enables threads dumping on failure 
(default value is true)
 - `dumpThreadsThrottlingTimeoutMillis` Timeout for throttling of thread dumps 
generation. The default value is 10 sec

Also, it makes sense to pre-allocate a buffer at the start of the failure 
processor to increase the chances of OOM handling.
The size of this buffer should be configured as well. The default value is 16Kb.
 - `oomBufferSizeBites` Amount of memory reserved in the heap at node start in 
kilobytes, which can be dropped to increase the chances of success when 
handling OutOfMemoryError

  was:
The failure handler should be able to create and log a thread dump on failure.

Also, the FH configuration should be extended with the following fields:

 - `dumpThreadsOnFailure` This property enables threads dumping on failure 
(default value is true)
 - `dumpThreadsThrottlingTimeoutMillis` Timeout for throttling of thread dumps 
generation. The default value is 10 sec

Also, it makes sense to pre-allocate a buffer at the start of failure processor 
to increase 

_java.lang.OutOfMemoryError_ should be handled according to the provided 
failure handle.
All critical workers should handle failures as CRITICAL _WORKER_TERMINATED.

Some memory should be reserved at node start to increase the chances of OOM 
handling.

Definition of done:
 - _OomExceptionHandler_ implemented
 - This handler integrated into _NamedThreadFactory_


> Failure handler should support generating thread dump on failure
> ----------------------------------------------------------------
>
>                 Key: IGNITE-20449
>                 URL: https://issues.apache.org/jira/browse/IGNITE-20449
>             Project: Ignite
>          Issue Type: Improvement
>            Reporter: Vyacheslav Koptilin
>            Assignee: Vyacheslav Koptilin
>            Priority: Major
>              Labels: ignite-3
>
> The failure handler should be able to create and log a thread dump on failure.
> The failure processor configuration should be extended with the following 
> fields:
>  - `dumpThreadsOnFailure` This property enables threads dumping on failure 
> (default value is true)
>  - `dumpThreadsThrottlingTimeoutMillis` Timeout for throttling of thread 
> dumps generation. The default value is 10 sec
> Also, it makes sense to pre-allocate a buffer at the start of the failure 
> processor to increase the chances of OOM handling.
> The size of this buffer should be configured as well. The default value is 
> 16Kb.
>  - `oomBufferSizeBites` Amount of memory reserved in the heap at node start 
> in kilobytes, which can be dropped to increase the chances of success when 
> handling OutOfMemoryError



--
This message was sent by Atlassian Jira
(v8.20.10#820010)

Reply via email to