[
https://issues.apache.org/jira/browse/IGNITE-25212?page=com.atlassian.jira.plugin.system.issuetabpanels:all-tabpanel
]
Ivan Bessonov updated IGNITE-25212:
-----------------------------------
Description:
In watch processor, we read the previous value of every updated key. This read
should not fail, even if the previous value is already compacted tombstone.
Currently, it might fail if we had a race in compactor, the left us some links
to already deleted revisions.
was:
Last two checks in this test:
{code:java}
AbstractCompactionKeyValueStorageTest#testGetSingleEntryAndCompactionForBarKey
{code}
We can return a tombstone instead of throwing an exception, if we encountered
it. (or an empty value in other similar cases, it tombstone timestamp is not
available anymore)
> CompactedException when encountering a tombstone
> ------------------------------------------------
>
> Key: IGNITE-25212
> URL: https://issues.apache.org/jira/browse/IGNITE-25212
> Project: Ignite
> Issue Type: Bug
> Reporter: Ivan Bessonov
> Assignee: Ivan Bessonov
> Priority: Critical
> Labels: ignite-3
>
> In watch processor, we read the previous value of every updated key. This
> read should not fail, even if the previous value is already compacted
> tombstone.
> Currently, it might fail if we had a race in compactor, the left us some
> links to already deleted revisions.
--
This message was sent by Atlassian Jira
(v8.20.10#820010)
