3:29 a.m.
FLUSH: problems with TCP and concurrent startup/shutdowns
---------------------------------------------------------
Key: JGRP-603
URL: http://jira.jboss.com/jira/browse/JGRP-603
Project: JGroups
Issue Type: Bug
Reporter: Bela Ban
Assigned To: Vladimir Blagojevic
The attached ZIP file has code that reproduces this.
Modify props.props (pf.cluster.transport.protocol=udp) to "tcp" and
"tcp-nio" if you want to test the different stacks.
To reproduce:
- Start a number of instances (e.g. 5) concurrently. This almost never works, even under
"udp". Joiners' JOIN requests time out and they have to retry (possibly
because the coord is busy with the FLUSH protocol). They become singleton members and
*never* merge !
- This works fine without FLUSH
- With "udp", it works almost always, with "tcp" it works 50% of the
time, with "tcp-nio" is almost never works
- Randomly kill and restart instances
--
This message is automatically generated by JIRA.
-
If you think it was sent incorrectly contact one of the administrators:
http://jira.jboss.com/jira/secure/Administrators.jspa
-
For more information on JIRA, see: http://www.atlassian.com/software/jira
3:31 a.m.
New subject: [JBoss JIRA] Updated: (JGRP-603) FLUSH: problems with TCP and concurrent startup/shutdowns
[ http://jira.jboss.com/jira/browse/JGRP-603?page=all ]
Bela Ban updated JGRP-603:
--------------------------
Attachment: jgtest.zip
FLUSH: problems with TCP and concurrent startup/shutdowns
---------------------------------------------------------
Key: JGRP-603
URL: http://jira.jboss.com/jira/browse/JGRP-603
Project: JGroups
Issue Type: Bug
Reporter: Bela Ban
Assigned To: Vladimir Blagojevic
Attachments: jgtest.zip
The attached ZIP file has code that reproduces this.
Modify props.props (pf.cluster.transport.protocol=udp) to "tcp" and
"tcp-nio" if you want to test the different stacks.
To reproduce:
- Start a number of instances (e.g. 5) concurrently. This almost never works, even under
"udp". Joiners' JOIN requests time out and they have to retry (possibly
because the coord is busy with the FLUSH protocol). They become singleton members and
*never* merge !
- This works fine without FLUSH
- With "udp", it works almost always, with "tcp" it works 50% of the
time, with "tcp-nio" is almost never works
- Randomly kill and restart instances
--
This message is automatically generated by JIRA.
-
If you think it was sent incorrectly contact one of the administrators:
http://jira.jboss.com/jira/secure/Administrators.jspa
-
For more information on JIRA, see: http://www.atlassian.com/software/jira
3:33 a.m.
New subject: [JBoss JIRA] Updated: (JGRP-603) FLUSH: problems with TCP and concurrent startup/shutdowns
[ http://jira.jboss.com/jira/browse/JGRP-603?page=all ]
Bela Ban updated JGRP-603:
--------------------------
Fix Version/s: 2.6
Priority: Critical (was: Major)
FLUSH: problems with TCP and concurrent startup/shutdowns
---------------------------------------------------------
Key: JGRP-603
URL: http://jira.jboss.com/jira/browse/JGRP-603
Project: JGroups
Issue Type: Bug
Reporter: Bela Ban
Assigned To: Vladimir Blagojevic
Priority: Critical
Fix For: 2.6
Attachments: jgtest.zip
The attached ZIP file has code that reproduces this.
Modify props.props (pf.cluster.transport.protocol=udp) to "tcp" and
"tcp-nio" if you want to test the different stacks.
To reproduce:
- Start a number of instances (e.g. 5) concurrently. This almost never works, even under
"udp". Joiners' JOIN requests time out and they have to retry (possibly
because the coord is busy with the FLUSH protocol). They become singleton members and
*never* merge !
- This works fine without FLUSH
- With "udp", it works almost always, with "tcp" it works 50% of the
time, with "tcp-nio" is almost never works
- Randomly kill and restart instances
--
This message is automatically generated by JIRA.
-
If you think it was sent incorrectly contact one of the administrators:
http://jira.jboss.com/jira/secure/Administrators.jspa
-
For more information on JIRA, see: http://www.atlassian.com/software/jira
3:43 a.m.
New subject: [JBoss JIRA] Commented: (JGRP-603) FLUSH: problems with TCP and concurrent startup/shutdowns
[ http://jira.jboss.com/jira/browse/JGRP-603?page=comments#action_12380083 ]
Bela Ban commented on JGRP-603:
-------------------------------
Actually, this works really well with "udp" (I probably tested "tcp"
when I thought I was testing "udp")...
But there are definitely issues with "tcp". Most issues are with
"tcp-nio", and we should look into those, although tcp-nio has not yet been
really well tested (and supported).
FLUSH: problems with TCP and concurrent startup/shutdowns
---------------------------------------------------------
Key: JGRP-603
URL: http://jira.jboss.com/jira/browse/JGRP-603
Project: JGroups
Issue Type: Bug
Reporter: Bela Ban
Assigned To: Vladimir Blagojevic
Priority: Critical
Fix For: 2.6
Attachments: jgtest.zip
The attached ZIP file has code that reproduces this.
Modify props.props (pf.cluster.transport.protocol=udp) to "tcp" and
"tcp-nio" if you want to test the different stacks.
To reproduce:
- Start a number of instances (e.g. 5) concurrently. This almost never works, even under
"udp". Joiners' JOIN requests time out and they have to retry (possibly
because the coord is busy with the FLUSH protocol). They become singleton members and
*never* merge !
- This works fine without FLUSH
- With "udp", it works almost always, with "tcp" it works 50% of the
time, with "tcp-nio" is almost never works
- Randomly kill and restart instances
--
This message is automatically generated by JIRA.
-
If you think it was sent incorrectly contact one of the administrators:
http://jira.jboss.com/jira/secure/Administrators.jspa
-
For more information on JIRA, see: http://www.atlassian.com/software/jira
4:02 a.m.
New subject: [JBoss JIRA] Commented: (JGRP-603) FLUSH: problems with TCP and concurrent startup/shutdowns
[ http://jira.jboss.com/jira/browse/JGRP-603?page=comments#action_12380086 ]
Bela Ban commented on JGRP-603:
-------------------------------
Okay, tcp and tcp-nio now work too if the following changes are made:
- Remove UNICAST from the tcp and tcp-nio stacks
- Set suspect_on_send_failure="true" in TCP and TCP_NIO
FLUSH: problems with TCP and concurrent startup/shutdowns
---------------------------------------------------------
Key: JGRP-603
URL: http://jira.jboss.com/jira/browse/JGRP-603
Project: JGroups
Issue Type: Bug
Reporter: Bela Ban
Assigned To: Vladimir Blagojevic
Priority: Critical
Fix For: 2.6
Attachments: jgtest.zip
The attached ZIP file has code that reproduces this.
Modify props.props (pf.cluster.transport.protocol=udp) to "tcp" and
"tcp-nio" if you want to test the different stacks.
To reproduce:
- Start a number of instances (e.g. 5) concurrently. This almost never works, even under
"udp". Joiners' JOIN requests time out and they have to retry (possibly
because the coord is busy with the FLUSH protocol). They become singleton members and
*never* merge !
- This works fine without FLUSH
- With "udp", it works almost always, with "tcp" it works 50% of the
time, with "tcp-nio" is almost never works
- Randomly kill and restart instances
--
This message is automatically generated by JIRA.
-
If you think it was sent incorrectly contact one of the administrators:
http://jira.jboss.com/jira/secure/Administrators.jspa
-
For more information on JIRA, see: http://www.atlassian.com/software/jira
4:33 p.m.
New subject: [JBoss JIRA] Updated: (JGRP-603) FLUSH: problems with TCP and concurrent startup/shutdowns
[ http://jira.jboss.com/jira/browse/JGRP-603?page=all ]
Bela Ban updated JGRP-603:
--------------------------
Attachment: 2.txt
FLUSH: problems with TCP and concurrent startup/shutdowns
---------------------------------------------------------
Key: JGRP-603
URL: http://jira.jboss.com/jira/browse/JGRP-603
Project: JGroups
Issue Type: Bug
Reporter: Bela Ban
Assigned To: Vladimir Blagojevic
Priority: Critical
Fix For: 2.6
Attachments: 1.txt, 2.txt, 3.txt, jgtest.zip
The attached ZIP file has code that reproduces this.
Modify props.props (pf.cluster.transport.protocol=udp) to "tcp" and
"tcp-nio" if you want to test the different stacks.
To reproduce:
- Start a number of instances (e.g. 5) concurrently. This almost never works, even under
"udp". Joiners' JOIN requests time out and they have to retry (possibly
because the coord is busy with the FLUSH protocol). They become singleton members and
*never* merge !
- This works fine without FLUSH
- With "udp", it works almost always, with "tcp" it works 50% of the
time, with "tcp-nio" is almost never works
- Randomly kill and restart instances
--
This message is automatically generated by JIRA.
-
If you think it was sent incorrectly contact one of the administrators:
http://jira.jboss.com/jira/secure/Administrators.jspa
-
For more information on JIRA, see: http://www.atlassian.com/software/jira
4:33 p.m.
New subject: [JBoss JIRA] Updated: (JGRP-603) FLUSH: problems with TCP and concurrent startup/shutdowns
[ http://jira.jboss.com/jira/browse/JGRP-603?page=all ]
Bela Ban updated JGRP-603:
--------------------------
Attachment: 3.txt
FLUSH: problems with TCP and concurrent startup/shutdowns
---------------------------------------------------------
Key: JGRP-603
URL: http://jira.jboss.com/jira/browse/JGRP-603
Project: JGroups
Issue Type: Bug
Reporter: Bela Ban
Assigned To: Vladimir Blagojevic
Priority: Critical
Fix For: 2.6
Attachments: 1.txt, 2.txt, 3.txt, jgtest.zip
The attached ZIP file has code that reproduces this.
Modify props.props (pf.cluster.transport.protocol=udp) to "tcp" and
"tcp-nio" if you want to test the different stacks.
To reproduce:
- Start a number of instances (e.g. 5) concurrently. This almost never works, even under
"udp". Joiners' JOIN requests time out and they have to retry (possibly
because the coord is busy with the FLUSH protocol). They become singleton members and
*never* merge !
- This works fine without FLUSH
- With "udp", it works almost always, with "tcp" it works 50% of the
time, with "tcp-nio" is almost never works
- Randomly kill and restart instances
--
This message is automatically generated by JIRA.
-
If you think it was sent incorrectly contact one of the administrators:
http://jira.jboss.com/jira/secure/Administrators.jspa
-
For more information on JIRA, see: http://www.atlassian.com/software/jira
4:33 p.m.
New subject: [JBoss JIRA] Updated: (JGRP-603) FLUSH: problems with TCP and concurrent startup/shutdowns
[ http://jira.jboss.com/jira/browse/JGRP-603?page=all ]
Bela Ban updated JGRP-603:
--------------------------
Attachment: 1.txt
FLUSH: problems with TCP and concurrent startup/shutdowns
---------------------------------------------------------
Key: JGRP-603
URL: http://jira.jboss.com/jira/browse/JGRP-603
Project: JGroups
Issue Type: Bug
Reporter: Bela Ban
Assigned To: Vladimir Blagojevic
Priority: Critical
Fix For: 2.6
Attachments: 1.txt, 2.txt, 3.txt, jgtest.zip
The attached ZIP file has code that reproduces this.
Modify props.props (pf.cluster.transport.protocol=udp) to "tcp" and
"tcp-nio" if you want to test the different stacks.
To reproduce:
- Start a number of instances (e.g. 5) concurrently. This almost never works, even under
"udp". Joiners' JOIN requests time out and they have to retry (possibly
because the coord is busy with the FLUSH protocol). They become singleton members and
*never* merge !
- This works fine without FLUSH
- With "udp", it works almost always, with "tcp" it works 50% of the
time, with "tcp-nio" is almost never works
- Randomly kill and restart instances
--
This message is automatically generated by JIRA.
-
If you think it was sent incorrectly contact one of the administrators:
http://jira.jboss.com/jira/secure/Administrators.jspa
-
For more information on JIRA, see: http://www.atlassian.com/software/jira
3:42 p.m.
New subject: [JBoss JIRA] Updated: (JGRP-603) FLUSH: problems with TCP and concurrent startup/shutdowns
[ http://jira.jboss.com/jira/browse/JGRP-603?page=all ]
Bela Ban updated JGRP-603:
--------------------------
Attachment: 103.txt
FLUSH: problems with TCP and concurrent startup/shutdowns
---------------------------------------------------------
Key: JGRP-603
URL: http://jira.jboss.com/jira/browse/JGRP-603
Project: JGroups
Issue Type: Bug
Reporter: Bela Ban
Assigned To: Vladimir Blagojevic
Priority: Critical
Fix For: 2.6
Attachments: 1.txt, 103.txt, 104.txt, 105.txt, 2.txt, 3.txt, jgtest.zip
The attached ZIP file has code that reproduces this.
Modify props.props (pf.cluster.transport.protocol=udp) to "tcp" and
"tcp-nio" if you want to test the different stacks.
To reproduce:
- Start a number of instances (e.g. 5) concurrently. This almost never works, even under
"udp". Joiners' JOIN requests time out and they have to retry (possibly
because the coord is busy with the FLUSH protocol). They become singleton members and
*never* merge !
- This works fine without FLUSH
- With "udp", it works almost always, with "tcp" it works 50% of the
time, with "tcp-nio" is almost never works
- Randomly kill and restart instances
--
This message is automatically generated by JIRA.
-
If you think it was sent incorrectly contact one of the administrators:
http://jira.jboss.com/jira/secure/Administrators.jspa
-
For more information on JIRA, see: http://www.atlassian.com/software/jira
3:42 p.m.
New subject: [JBoss JIRA] Updated: (JGRP-603) FLUSH: problems with TCP and concurrent startup/shutdowns
[ http://jira.jboss.com/jira/browse/JGRP-603?page=all ]
Bela Ban updated JGRP-603:
--------------------------
Attachment: 104.txt
FLUSH: problems with TCP and concurrent startup/shutdowns
---------------------------------------------------------
Key: JGRP-603
URL: http://jira.jboss.com/jira/browse/JGRP-603
Project: JGroups
Issue Type: Bug
Reporter: Bela Ban
Assigned To: Vladimir Blagojevic
Priority: Critical
Fix For: 2.6
Attachments: 1.txt, 103.txt, 104.txt, 105.txt, 2.txt, 3.txt, jgtest.zip
The attached ZIP file has code that reproduces this.
Modify props.props (pf.cluster.transport.protocol=udp) to "tcp" and
"tcp-nio" if you want to test the different stacks.
To reproduce:
- Start a number of instances (e.g. 5) concurrently. This almost never works, even under
"udp". Joiners' JOIN requests time out and they have to retry (possibly
because the coord is busy with the FLUSH protocol). They become singleton members and
*never* merge !
- This works fine without FLUSH
- With "udp", it works almost always, with "tcp" it works 50% of the
time, with "tcp-nio" is almost never works
- Randomly kill and restart instances
--
This message is automatically generated by JIRA.
-
If you think it was sent incorrectly contact one of the administrators:
http://jira.jboss.com/jira/secure/Administrators.jspa
-
For more information on JIRA, see: http://www.atlassian.com/software/jira
3:42 p.m.
New subject: [JBoss JIRA] Updated: (JGRP-603) FLUSH: problems with TCP and concurrent startup/shutdowns
[ http://jira.jboss.com/jira/browse/JGRP-603?page=all ]
Bela Ban updated JGRP-603:
--------------------------
Attachment: 105.txt
FLUSH: problems with TCP and concurrent startup/shutdowns
---------------------------------------------------------
Key: JGRP-603
URL: http://jira.jboss.com/jira/browse/JGRP-603
Project: JGroups
Issue Type: Bug
Reporter: Bela Ban
Assigned To: Vladimir Blagojevic
Priority: Critical
Fix For: 2.6
Attachments: 1.txt, 103.txt, 104.txt, 105.txt, 2.txt, 3.txt, jgtest.zip
The attached ZIP file has code that reproduces this.
Modify props.props (pf.cluster.transport.protocol=udp) to "tcp" and
"tcp-nio" if you want to test the different stacks.
To reproduce:
- Start a number of instances (e.g. 5) concurrently. This almost never works, even under
"udp". Joiners' JOIN requests time out and they have to retry (possibly
because the coord is busy with the FLUSH protocol). They become singleton members and
*never* merge !
- This works fine without FLUSH
- With "udp", it works almost always, with "tcp" it works 50% of the
time, with "tcp-nio" is almost never works
- Randomly kill and restart instances
--
This message is automatically generated by JIRA.
-
If you think it was sent incorrectly contact one of the administrators:
http://jira.jboss.com/jira/secure/Administrators.jspa
-
For more information on JIRA, see: http://www.atlassian.com/software/jira
3:52 p.m.
New subject: [JBoss JIRA] Commented: (JGRP-603) FLUSH: problems with TCP and concurrent startup/shutdowns
[ http://jira.jboss.com/jira/browse/JGRP-603?page=comments#action_12382877 ]
Vladimir Blagojevic commented on JGRP-603:
------------------------------------------
Bela described the scenario to reproduce this issue:
- View is V2 {A,B,C}
- Member C has address C:7602
- Member C is killed and *immediately* restarted (before A can exclude it)
- Member C has the *same* address: C:7602
- On JOIN, A will return the *existing view* V2 because it sees that C is still a member
!
Does C's FLUSH.down() unblock in such a case ? I don't think so, it only unblocks
on reception of the STOP-FLUSH message ! However, this message is never received because
the coord (A) never even starts the flush in this case !
FLUSH: problems with TCP and concurrent startup/shutdowns
---------------------------------------------------------
Key: JGRP-603
URL: http://jira.jboss.com/jira/browse/JGRP-603
Project: JGroups
Issue Type: Bug
Reporter: Bela Ban
Assigned To: Vladimir Blagojevic
Priority: Critical
Fix For: 2.6
Attachments: 1.txt, 103.txt, 104.txt, 105.txt, 2.txt, 3.txt, jgtest.zip
The attached ZIP file has code that reproduces this.
Modify props.props (pf.cluster.transport.protocol=udp) to "tcp" and
"tcp-nio" if you want to test the different stacks.
To reproduce:
- Start a number of instances (e.g. 5) concurrently. This almost never works, even under
"udp". Joiners' JOIN requests time out and they have to retry (possibly
because the coord is busy with the FLUSH protocol). They become singleton members and
*never* merge !
- This works fine without FLUSH
- With "udp", it works almost always, with "tcp" it works 50% of the
time, with "tcp-nio" is almost never works
- Randomly kill and restart instances
--
This message is automatically generated by JIRA.
-
If you think it was sent incorrectly contact one of the administrators:
http://jira.jboss.com/jira/secure/Administrators.jspa
-
For more information on JIRA, see: http://www.atlassian.com/software/jira
8:29 a.m.
New subject: [JBoss JIRA] Commented: (JGRP-603) FLUSH: problems with TCP and concurrent startup/shutdowns
[ http://jira.jboss.com/jira/browse/JGRP-603?page=comments#action_12386176 ]
Bela Ban commented on JGRP-603:
-------------------------------
Solved by preventing the reincarnation issue
FLUSH: problems with TCP and concurrent startup/shutdowns
---------------------------------------------------------
Key: JGRP-603
URL: http://jira.jboss.com/jira/browse/JGRP-603
Project: JGroups
Issue Type: Bug
Reporter: Bela Ban
Assigned To: Vladimir Blagojevic
Priority: Critical
Fix For: 2.6
Attachments: 1.txt, 103.txt, 104.txt, 105.txt, 2.txt, 3.txt, jgtest.zip
The attached ZIP file has code that reproduces this.
Modify props.props (pf.cluster.transport.protocol=udp) to "tcp" and
"tcp-nio" if you want to test the different stacks.
To reproduce:
- Start a number of instances (e.g. 5) concurrently. This almost never works, even under
"udp". Joiners' JOIN requests time out and they have to retry (possibly
because the coord is busy with the FLUSH protocol). They become singleton members and
*never* merge !
- This works fine without FLUSH
- With "udp", it works almost always, with "tcp" it works 50% of the
time, with "tcp-nio" is almost never works
- Randomly kill and restart instances
--
This message is automatically generated by JIRA.
-
If you think it was sent incorrectly contact one of the administrators:
http://jira.jboss.com/jira/secure/Administrators.jspa
-
For more information on JIRA, see: http://www.atlassian.com/software/jira
8:29 a.m.
New subject: [JBoss JIRA] Resolved: (JGRP-603) FLUSH: problems with TCP and concurrent startup/shutdowns
[ http://jira.jboss.com/jira/browse/JGRP-603?page=all ]
Bela Ban resolved JGRP-603.
---------------------------
Resolution: Done
Solved by http://jira.jboss.com/jira/browse/JGRP-130
FLUSH: problems with TCP and concurrent startup/shutdowns
---------------------------------------------------------
Key: JGRP-603
URL: http://jira.jboss.com/jira/browse/JGRP-603
Project: JGroups
Issue Type: Bug
Reporter: Bela Ban
Assigned To: Vladimir Blagojevic
Priority: Critical
Fix For: 2.6
Attachments: 1.txt, 103.txt, 104.txt, 105.txt, 2.txt, 3.txt, jgtest.zip
The attached ZIP file has code that reproduces this.
Modify props.props (pf.cluster.transport.protocol=udp) to "tcp" and
"tcp-nio" if you want to test the different stacks.
To reproduce:
- Start a number of instances (e.g. 5) concurrently. This almost never works, even under
"udp". Joiners' JOIN requests time out and they have to retry (possibly
because the coord is busy with the FLUSH protocol). They become singleton members and
*never* merge !
- This works fine without FLUSH
- With "udp", it works almost always, with "tcp" it works 50% of the
time, with "tcp-nio" is almost never works
- Randomly kill and restart instances
--
This message is automatically generated by JIRA.
-
If you think it was sent incorrectly contact one of the administrators:
http://jira.jboss.com/jira/secure/Administrators.jspa
-
For more information on JIRA, see: http://www.atlassian.com/software/jira
6262
days inactive
6296
days old
13 comments
2 participants
participants (2)
-
Bela Ban (JIRA) -
Vladimir Blagojevic (JIRA)