[
https://issues.apache.org/jira/browse/MINIFI-384?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=16132132#comment-16132132
]
marco polo commented on MINIFI-384:
-----------------------------------
Timeline of events
1) FlowFile object created with stored = false
2) Connection of ProcA -> SiteToSite fails, thus calling rollback. FF is stored
in repo, but stored = false because the put on the connection never occurs
3) ResourceClaim count == 0, and check is performed
[here|https://github.com/apache/nifi-minifi-cpp/blob/rel/minifi-cpp-0.2.0/libminifi/src/FlowFileRecord.cpp#L112]
.
4) SiteToSite begins again after other MiNiFI-380 subtasks are resolved, thus
causing normal sequence to occur. Content is deleted despite being in flow file
repo
5) SiteToSite rolls back the session because content is deleted, thus causing
iteration over potentially hundreds of thousands of mis-deleted content claims.
Fix is to change Line 112 of FlowFileRecord.cpp so that we don't rely on the
stored variable and instead rely solely on the flow file repo.
> Site To Site protocol re-establishment when fixed loses state in the
> connection
> -------------------------------------------------------------------------------
>
> Key: MINIFI-384
> URL: https://issues.apache.org/jira/browse/MINIFI-384
> Project: Apache NiFi MiNiFi
> Issue Type: Sub-task
> Components: C++
> Reporter: marco polo
> Assignee: marco polo
>
> Protocol re-establishment should reset state. MINIFI-383 is somehow impacting
> this such that we can never continue processes because we're constantly
> rolling back sessions. This is currently being investigated and will be
> updated shortly.
--
This message was sent by Atlassian JIRA
(v6.4.14#64029)
