[
https://issues.apache.org/jira/browse/IGNITE-24470?page=com.atlassian.jira.plugin.system.issuetabpanels:all-tabpanel
]
Evgeny Stanilovsky updated IGNITE-24470:
----------------------------------------
Description:
Nodes observes meta-store globalOperationStatus#STARTED and start to fill node
related information - CpccOperation#operations
Nodes self update meta-store status records according to planned resizing jobs.
Need to correctly handle primary replica change, it can be three variants here:
# CPCC is just started, globalOperationStatus in STARTED state and no node is
assigned primary changed partition - further cpcc related flow can be continued.
# CPCC in progress and primary replica change affects already processing
partitions - cpcc need to fail (with further zone unlock and garbage clean)
with appropriate return code and CpccOperation#failureReason
# CPCC in progress but primary replica change happens after partition already
replicated into newly created group - further cpcc related flow can be
continued.
Definition of done: upper cases need to be covered by tests, related
functionality implemented.
was:
Nodes observes meta-store globalOperationStatus#STARTED and start to fill node
related information - CpccOperation#operations
Nodes self update meta-store status records according to planned resizing jobs.
Need to correctly handle primary replica change, it can be three variants here:
# CPCC is just started, globalOperationStatus in (INITIATED or STARTED) states
and no node is assigned primary changed partition - further cpcc related flow
can be continued.
# CPCC in progress and primary replica change affects already processing
partitions - cpcc need to fail (with further zone unlock and garbage clean)
with appropriate return code and CpccOperation#failureReason
# CPCC in progress but primary replica change happens after partition already
replicated into newly created group - further cpcc related flow can be
continued.
Definition of done: upper cases need to be covered by tests, related
functionality implemented.
> CPCC. Handling primary replica change
> -------------------------------------
>
> Key: IGNITE-24470
> URL: https://issues.apache.org/jira/browse/IGNITE-24470
> Project: Ignite
> Issue Type: Task
> Components: sql
> Affects Versions: 3.0
> Reporter: Evgeny Stanilovsky
> Priority: Major
> Labels: ignite-3
>
> Nodes observes meta-store globalOperationStatus#STARTED and start to fill
> node related information - CpccOperation#operations
> Nodes self update meta-store status records according to planned resizing
> jobs.
> Need to correctly handle primary replica change, it can be three variants
> here:
> # CPCC is just started, globalOperationStatus in STARTED state and no node is
> assigned primary changed partition - further cpcc related flow can be
> continued.
> # CPCC in progress and primary replica change affects already processing
> partitions - cpcc need to fail (with further zone unlock and garbage clean)
> with appropriate return code and CpccOperation#failureReason
> # CPCC in progress but primary replica change happens after partition already
> replicated into newly created group - further cpcc related flow can be
> continued.
> Definition of done: upper cases need to be covered by tests, related
> functionality implemented.
--
This message was sent by Atlassian Jira
(v8.20.10#820010)
