Bug 745880 (EDG-53)

Summary:	memcached commands noreply option error
Product:	[JBoss] JBoss Data Grid 5	Reporter:	Michal Linhard <mlinhard>
Component:	Infinispan	Assignee:	Default User <jbpapp-maint>
Status:	CLOSED NEXTRELEASE	QA Contact:
Severity:	high	Docs Contact:
Priority:	high
Version:	EAP 5.1.0 EDG TP	CC:	galder.zamarreno, mlinhard, nobody
Target Milestone:	---
Target Release:	EAP 5.1.0 EDG TP
Hardware:	Unspecified
OS:	Unspecified
URL:	http://jira.jboss.org/jira/browse/EDG-53
Whiteboard:
Fixed In Version:		Doc Type:	Bug Fix
Doc Text:		Story Points:	---
Clone Of:		Environment:
Last Closed:	2011-05-09 12:23:14 UTC	Type:	Bug
Regression:	---	Mount Type:	---
Documentation:	---	CRM:
Verified Versions:		Category:	---
oVirt Team:	---	RHEL 7.3 requirements from Atomic Host:
Cloudforms Team:	---	Target Upstream Version:

Description Michal Linhard 2011-04-28 09:26:50 UTC

project_key: EDG

there was a regression:
https://hudson.qa.jboss.com/hudson/view/EDG/job/edg-51x-build-alt/41/testReport/junit/org.jboss.test.cluster.datagrid.test/
(testcases testNoReply*)

for the infinispan snapshot version: 4.2.2-SNAPSHOT-20110418.174252
edg build: http://hudson.qa.jboss.com/hudson/view/EDG/job/edg-51x-build-fast/16/artifact/edg-ispn-4.2.2-SNAPSHOT.zip

I'm working on the steps to reproduce

Comment 1 Michal Linhard 2011-04-28 13:25:06 UTC

this might be related:
echo -e "delete a noreply\r" | nc localhost 11211
produces
SERVER_ERROR java.lang.NullPointerException

Comment 2 Michal Linhard 2011-04-28 14:52:26 UTC

it seems as delete command inherits noreply flag from previous command with noreply:
{code}
$ echo -e "set a 0 0 1\r\na\r\ndelete a\r\ndelete a\r\ndelete a\r\ndelete a\r\n" | nc localhost 11211
STORED
DELETED
NOT_FOUND
NOT_FOUND
NOT_FOUND
$ echo -e "set a 0 0 1 noreply\r\na\r\ndelete a\r\ndelete a\r\ndelete a\r\ndelete a\r\n" | nc localhost 11211
$
{code}

Comment 3 Michal Linhard 2011-04-28 14:53:27 UTC

and therefore all testNoReply* tests get read timeout because they expect some reply from delete command in tearDown().

Comment 4 Galder Zamarreño 2011-04-29 15:24:38 UTC

This could well be a regression as a result of the changes I made in the decoder to reduce memory consumption. I had to change quite a few things around. I'll create an ISPN jira to investigate this.

Comment 5 Galder Zamarreño 2011-04-29 15:30:44 UTC

Link: Added: This issue depends ISPN-1083

Comment 6 Galder Zamarreño 2011-05-03 06:45:20 UTC

Michal, I solved this yesterday while travelling back to Switzerland. In fact there're two issues but I've just bundled them into one. As you rightly guessed, on one side old parameters were being leaked into next request if the latter one did not have any extra params. On top of that there was the NPE with no reply. I've now added a listener that verifies that after a noreply op, the operation in the cache has been executed.

Comment 7 Michal Linhard 2011-05-11 08:31:24 UTC

The new snapshot 4.2.2-20110509.161652 solved the problem:
http://hudson.qa.jboss.com/hudson/view/EDG/job/edg-51x-build-alt/53/testReport/
that existed in previous run:
http://hudson.qa.jboss.com/hudson/view/EDG/job/edg-51x-build-alt/52/testReport/

Comment 8 Anne-Louise Tangring 2011-10-11 17:05:54 UTC

Docs QE Status: Removed: NEW