[
https://issues.apache.org/jira/browse/IGNITE-21213?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=17809297#comment-17809297
]
Denis Chudov edited comment on IGNITE-21213 at 1/22/24 8:32 AM:
----------------------------------------------------------------
There are 3 aspects of this problem:
# Handling the events processing should be done within the replica stripe,
will be covered in IGNITE-21258
# We shouldn't allow to process transactional requests and events handling
simultaneously wrt async calls, so we should make something like read-write
lock (but not literally, because simple read-write lock will lead to
deadlocks). Can be done in this ticket.
# We shouldn't allow to process new transactional requests until the event
handling is done, PlacementDriver#previousPrimaryExpired future can be used for
this: IGNITE-21325
was (Author: denis chudov):
There are 3 aspects of this problem:
# Handling the events processing should be done within the replica stripe,
will be covered in IGNITE-21258
# We shouldn't allow to process transactional requests and events handling
simultaneously wrt async calls, so we should make something like read-write
lock (but not literally, because simple read-write lock will lead to
deadlocks). Can be done in this ticket.
# We shouldn't allow to process new transactional requests until the event
handling is done, PlacementDriver#previousPrimaryExpired future can be used for
this.
> Coordination of mechanisms of determination for primary on replicaside
> ----------------------------------------------------------------------
>
> Key: IGNITE-21213
> URL: https://issues.apache.org/jira/browse/IGNITE-21213
> Project: Ignite
> Issue Type: Bug
> Reporter: Vladislav Pyatkov
> Priority: Major
> Labels: ignite-3
>
> h3. Motivation
> In the replica listener, we have unconsidered mechanisms between each other
> to determine primary rteplica. The first one is based on the placement driver
> API (it is used in _PartitionReplicaListener#ensureReplicaIsPrimary_) and the
> other one is based on the placement driver events (the events are hadeled by
> two methods: _ReplicaManager#onPrimaryReplicaElected_,
> _ReplicaManager#onPrimaryReplicaExpired_).
> Because the replica messages and events are handled in different threads, any
> variety of processing is possible. For example, the replica can release all
> transaction locks (by
> PRIMARY_REPLICA_EXPIRED event) and then handle a message for this transaction
> (because ensureReplicaIsPrimary was done before), assuming that all the locks
> are holding.
> h3. Definition of done
> The two mechanisms work in coordination.
--
This message was sent by Atlassian Jira
(v8.20.10#820010)
