You are viewing a plain text version of this content. The canonical link for it is here.
Posted to issues@geode.apache.org by "Anilkumar Gingade (JIRA)" <ji...@apache.org> on 2018/09/14 19:51:00 UTC
[jira] [Reopened] (GEODE-5605) In flight cache operation may not be
part of GII if the GII provider cache closes
[ https://issues.apache.org/jira/browse/GEODE-5605?page=com.atlassian.jira.plugin.system.issuetabpanels:all-tabpanel ]
Anilkumar Gingade reopened GEODE-5605:
--------------------------------------
As part of fixing this issue, added change to address EntryNotFoundException found during retry of entry operation.
Scenario:
-- Destroy is performed by Partitioned Region Accessor
-- This gets applied on Node1; replicates it to Node3. Does not replicates to Node2.
-- Node1 cache closes; before sending message back to accessor.
-- Node2 and Node3 does region sync
-- Accessor retries the op on Node2. Since its already applied Node2, it throws EntryNotFoundException
But this breaks the scenario; where a remove is done for an entry which is not present; these ops could get retried; and may not throw expected EntryNotFoundException. This needs to be addressed in another ticket.
> In flight cache operation may not be part of GII if the GII provider cache closes
> ---------------------------------------------------------------------------------
>
> Key: GEODE-5605
> URL: https://issues.apache.org/jira/browse/GEODE-5605
> Project: Geode
> Issue Type: Improvement
> Components: regions
> Affects Versions: 1.6.0
> Reporter: Anilkumar Gingade
> Assignee: Anilkumar Gingade
> Priority: Major
> Labels: pull-request-available
> Fix For: 1.8.0
>
> Time Spent: 40m
> Remaining Estimate: 0h
>
> An in-flight operation may not be seen by the new GII requester if the GII provider cache closes. This happens when there are more than one replica for the region.
> Scenario:
> There are 3 nodes (N1 with primary bucket, N2 with secondary bucket and N3 new node doing GII request)
> * accessor begins putAll
> * accessor sends putAll to N1
> * N1 locally processes putAll
> * N1 attempts to replicate putAll to N2
> * during this putAll, N1 is shutdown normally
> * during this putAll (after N1 has started to shutdown), N3 attempts to GII from N1
> * it fails because N1 is shutting down
> * N3 then GIIs from N2 (but before the put is received by it)
> * after GII, N3 doesn't have the entry, but N2 does
> * accessor completes putAll successfully
>
--
This message was sent by Atlassian JIRA
(v7.6.3#76005)