[jboss-jira] [JBoss JIRA] Commented: (JGRP-678) NPE in NAKACK under startup load
[ http://jira.jboss.com/jira/browse/JGRP-678?page=comments#action_12396808 ]
Bela Ban commented on JGRP-678:
-------------------------------
I don't think creating the entry for local_addr in xmit_table when not found is the
right approach, for two reasons:
1. We should find out *why* local_addr was removed from xmit_table.
local_addr is added on the first view and should never be removed
! I'll create a JIRA to investigate
2. Indiscriminately adding local_addr if not found may actually be
dangerous: if we unilaterally removed local_addr (and I have to
find out why), and others still have their entry for us (e.g.
#25), then we will start at seqno #1 again, and our first 25
messages might get discarded !
I somewhat agree with the trace issue, but this information is really useful when such a
situation occurs. I'll probably add another flag and trace when set (default=on).
NPE in NAKACK under startup load
--------------------------------
Key: JGRP-678
URL: http://jira.jboss.com/jira/browse/JGRP-678
Project: JGroups
Issue Type: Bug
Reporter: Bela Ban
Assigned To: Bela Ban
Fix For: 2.7, 2.6.2
Attachments: NAKACK.patch.txt, NAKACK.stack.txt, peerProtocolConfig.xml
Here's another NPE that we found, this time in NAKACK in 2.6.1. Again, we
are starting up ~30 systems at once which causes some thrashing for a couple
of minutes until things settle down.
You will notice that we also moved over to trace the dumping of the NAKACK
table which at some points threatened to overwhelm our log systems. That
has (I think) been mentioned here before.
--
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