[
https://issues.apache.org/jira/browse/IGNITE-28095?page=com.atlassian.jira.plugin.system.issuetabpanels:all-tabpanel
]
Roman Puchkovskiy updated IGNITE-28095:
---------------------------------------
Description:
Metastorage (MS) safe time is advanced both by commands creating new revisions
and special SyncTimeCommand commands that do not create new revisions, but just
advance the safe time.
On node start, we recover MS safe time from latest revision currently applied
by the MS group. This means that we miss SyncTimeCommands executed after the
latest revision, and we can (and usually do) restore MS safe time to the value
preceding the actual value that was seen by the node. This breaks MS safe time
monotonicity.
To make it possible to restart MS safe time completely, we need to persist MS
safe time to MS KV storage on SyncTimeCommand handling.
> Recover exact MS safe time on node start
> ----------------------------------------
>
> Key: IGNITE-28095
> URL: https://issues.apache.org/jira/browse/IGNITE-28095
> Project: Ignite
> Issue Type: Improvement
> Reporter: Roman Puchkovskiy
> Priority: Major
> Labels: ignite-3
>
> Metastorage (MS) safe time is advanced both by commands creating new
> revisions and special SyncTimeCommand commands that do not create new
> revisions, but just advance the safe time.
> On node start, we recover MS safe time from latest revision currently applied
> by the MS group. This means that we miss SyncTimeCommands executed after the
> latest revision, and we can (and usually do) restore MS safe time to the
> value preceding the actual value that was seen by the node. This breaks MS
> safe time monotonicity.
> To make it possible to restart MS safe time completely, we need to persist MS
> safe time to MS KV storage on SyncTimeCommand handling.
--
This message was sent by Atlassian Jira
(v8.20.10#820010)
