[
https://issues.apache.org/jira/browse/IGNITE-26759?page=com.atlassian.jira.plugin.system.issuetabpanels:all-tabpanel
]
Mirza Aliev updated IGNITE-26759:
---------------------------------
Description:
h3. Motivation
Currently if a user tries to perform partition restart with clean up and there
is no enough alive nodes, operation will be rejected.
Enough nodes means that
{code:java}
aliveStableNodes.size() > (zoneDescriptor.replicas() / 2 + 1)
{code}
We can weaken this restriction with the special flag {{--force}} if user is ok
that such operation could lead to data loss.
h3. Definition of done
* {{--force}} flag is added to CLI and REST, so restart with clean up performed
even if there is no enough alive nodes with data
was:
h3. Motivation
Currently if a user tries to perform partition restart with clean up and there
is no enough alive nodes, operation will be rejected.
Enough nodes means that
{code:java}
aliveStableNodes.size() > (zoneDescriptor.replicas() / 2 + 1)
{code}
We can weaken this restriction with the special flag {{--force}} if user is ok
that such operation could lead to data loss.
h3. Definition of done
* {{--force}} flag is added to CLI and REST, so restart with clean up performed
even if there is no enough alive nodes with data
> Allow to do partition restart with clean up even if there is not enough nodes
> -----------------------------------------------------------------------------
>
> Key: IGNITE-26759
> URL: https://issues.apache.org/jira/browse/IGNITE-26759
> Project: Ignite
> Issue Type: Improvement
> Reporter: Mirza Aliev
> Priority: Major
> Labels: ignite-3
>
> h3. Motivation
> Currently if a user tries to perform partition restart with clean up and
> there is no enough alive nodes, operation will be rejected.
> Enough nodes means that
> {code:java}
> aliveStableNodes.size() > (zoneDescriptor.replicas() / 2 + 1)
> {code}
> We can weaken this restriction with the special flag {{--force}} if user is
> ok that such operation could lead to data loss.
> h3. Definition of done
> * {{--force}} flag is added to CLI and REST, so restart with clean up
> performed even if there is no enough alive nodes with data
--
This message was sent by Atlassian Jira
(v8.20.10#820010)
