[ 
https://issues.apache.org/jira/browse/IGNITE-21258?page=com.atlassian.jira.plugin.system.issuetabpanels:all-tabpanel
 ]

Vladislav Pyatkov updated IGNITE-21258:
---------------------------------------
    Description: 
h3. Motivation
We seek to decrease the load on the metastorage thread because when the MC 
thread is holding, it blocks the handling of other events.
{code}
placementDriver.listen(PrimaryReplicaEvent.PRIMARY_REPLICA_ELECTED, 
this::onPrimaryReplicaElected);
placementDriver.listen(PrimaryReplicaEvent.PRIMARY_REPLICA_EXPIRED, 
this::onPrimaryReplicaExpired);
{code}
Currently, both primary replica events are handled on the MC thread, though the 
striped executer is a well-suited place to do it.

h3. Definition of done
Move handling of the primary replica events (PRIMARY_REPLICA_ELECTED, 
PRIMARY_REPLICA_EXPIRED) in the striped pool.

  was:
h3. Motivation
We seek to decrease the load on the metastorage thread because when the MC 
thread is holding, it blocks the handling of other events.
{code}
placementDriver.listen(PrimaryReplicaEvent.PRIMARY_REPLICA_ELECTED, 
this::onPrimaryReplicaElected);
placementDriver.listen(PrimaryReplicaEvent.PRIMARY_REPLICA_EXPIRED, 
this::onPrimaryReplicaExpired);
{code}
Currently, both primary replica events are handled on the MC thread, though the 
striped executer is a well-suited place to do it.

h3. Defenition of done
Move handling of the primary replica events (PRIMARY_REPLICA_ELECTED, 
PRIMARY_REPLICA_EXPIRED) in the striped pool.


> Handle primary replica events in the striped pool
> -------------------------------------------------
>
>                 Key: IGNITE-21258
>                 URL: https://issues.apache.org/jira/browse/IGNITE-21258
>             Project: Ignite
>          Issue Type: Improvement
>            Reporter: Vladislav Pyatkov
>            Priority: Major
>              Labels: ignite-3
>
> h3. Motivation
> We seek to decrease the load on the metastorage thread because when the MC 
> thread is holding, it blocks the handling of other events.
> {code}
> placementDriver.listen(PrimaryReplicaEvent.PRIMARY_REPLICA_ELECTED, 
> this::onPrimaryReplicaElected);
> placementDriver.listen(PrimaryReplicaEvent.PRIMARY_REPLICA_EXPIRED, 
> this::onPrimaryReplicaExpired);
> {code}
> Currently, both primary replica events are handled on the MC thread, though 
> the striped executer is a well-suited place to do it.
> h3. Definition of done
> Move handling of the primary replica events (PRIMARY_REPLICA_ELECTED, 
> PRIMARY_REPLICA_EXPIRED) in the striped pool.



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

Reply via email to