[
https://issues.apache.org/jira/browse/IGNITE-16907?page=com.atlassian.jira.plugin.system.issuetabpanels:all-tabpanel
]
Alexander Lapin updated IGNITE-16907:
-------------------------------------
Description:
h4. Problem
Storage that is used in raft state machine, e.g. PartitionListener probably
won't flush every insert on disk on per-insert basis rather it will write
inserts into some sort of WAL and do checkpoints periodically. Such, WAL-based
solution, within the context of local recovery, allows to load data into raft
state machine starting from last processed appliedIndex + 1. So far so good,
but...
It worth mentioning, that we already have an entity called a Raft log that can
be used as a WAL to solve recovery problem mentioned above, which makes it
possible to eliminate in-storage WAL.
was:
h4. Problem
Storage that is used in raft state machine, e.g. PartitionListener probably
won't flush every insert on disk on per-insert basis rather it will write
inserts into some sort of WAL and do checkpoints periodically. Such, WAL-based
solution, within the context of local recovery, allows to continue loading data
into raft state machine starting from last processed appliedIndex + 1. So far
so good.
However, it worth mentioning, that we already have an entity called a Raft log
that can be used as a WAL to solve recovery problem mentioned above, which
makes it possible to eliminate in-storage WAL.
> Add ability to use Raft log as storage WAL wihtin the scope of local recovery
> -----------------------------------------------------------------------------
>
> Key: IGNITE-16907
> URL: https://issues.apache.org/jira/browse/IGNITE-16907
> Project: Ignite
> Issue Type: Improvement
> Reporter: Alexander Lapin
> Priority: Major
> Labels: ignite-3
>
> h4. Problem
> Storage that is used in raft state machine, e.g. PartitionListener probably
> won't flush every insert on disk on per-insert basis rather it will write
> inserts into some sort of WAL and do checkpoints periodically. Such,
> WAL-based solution, within the context of local recovery, allows to load data
> into raft state machine starting from last processed appliedIndex + 1. So far
> so good, but...
> It worth mentioning, that we already have an entity called a Raft log that
> can be used as a WAL to solve recovery problem mentioned above, which makes
> it possible to eliminate in-storage WAL.
--
This message was sent by Atlassian Jira
(v8.20.7#820007)
