[
https://issues.apache.org/jira/browse/IGNITE-26760?page=com.atlassian.jira.plugin.system.issuetabpanels:all-tabpanel
]
Denis Chudov updated IGNITE-26760:
----------------------------------
Description:
h3. Motivation
When user performs partition restart with clean up, it is possible, that node
on which such restart is supposed to be performed is stopped and restart is not
happened.
We should consider logic that will recover such intent that didn't happen
because node restart.
Also, the restart of multiple groups is not possible.
We should store restart intentions in meta storage for all partitions that have
not been restarted.
*Definition of done*
Restart intentions for each partition are stored in meta storage (there should
be a collection of partitions mapped to nodes, it should be stored in meta
storage with a new meta storage key). When user calls a partitions restart, the
group id is added to the collection. When partition is restarted, the group id
is removed from collection. \
This ticket is only about collection managing. The process improvements should
be done under IGNITE-27342 .
was:
h3. Motivation
When user performs partition restart with clean up, it is possible, that node
on which such restart is supposed to be performed is stopped and restart is not
happened.
We should consider logic that will recover such intent that didn't happen
because node restart.
Also, the restart of multiple groups is not possible.
We should store restart intentions in meta storage for all partitions that have
not been restarted.
*Definition of done*
Restart intentions for each partition are stored in meta storage (there should
be a collection of partitions mapped to nodes, it should be stored in meta
storage with a new meta storage key). When user calls a partitions restart, the
group id is added to the collection. When partition is restarted, the group id
is removed from collection.
> Implement recovery of partition restart with clean up on node restart.
> ----------------------------------------------------------------------
>
> Key: IGNITE-26760
> URL: https://issues.apache.org/jira/browse/IGNITE-26760
> Project: Ignite
> Issue Type: Improvement
> Reporter: Mirza Aliev
> Priority: Major
> Labels: ignite-3
>
> h3. Motivation
> When user performs partition restart with clean up, it is possible, that node
> on which such restart is supposed to be performed is stopped and restart is
> not happened.
> We should consider logic that will recover such intent that didn't happen
> because node restart.
> Also, the restart of multiple groups is not possible.
> We should store restart intentions in meta storage for all partitions that
> have not been restarted.
> *Definition of done*
> Restart intentions for each partition are stored in meta storage (there
> should be a collection of partitions mapped to nodes, it should be stored in
> meta storage with a new meta storage key). When user calls a partitions
> restart, the group id is added to the collection. When partition is
> restarted, the group id is removed from collection. \
> This ticket is only about collection managing. The process improvements
> should be done under IGNITE-27342 .
--
This message was sent by Atlassian Jira
(v8.20.10#820010)
