[ http://jira.jboss.com/jira/browse/JGRP-707?page=all ] Gray Watson updated JGRP-707: ----------------------------- Attachment: NAKACK.java.patch This patch fixes the NPE although it is important to note that race conditions in a lot of the logic for putIfAbsent mean that the first couple of packets for each address may not be recorded correctly. Something like the following logic needs to be in a ton of places in the code but this may not be necessary for statistics. StatsEntry entry=map.get(key); if(entry == null) { entry = new StatsEntry(); StatsEntry entry2 = map.putIfAbsent(key, entry); if (entry2 != null) entry = entry2; } entry.xmit_reqs+=req; .... -- 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

[ http://jira.jboss.com/jira/browse/JGRP-707?page=comments#action_12402378 ] Bela Ban commented on JGRP-707: ------------------------------- Fixed this with ConcurrentHashMap. However, for HashMaps, I didn't change it, e.g.: private static void updateStats(HashMap<Address,StatsEntry> map, Address key, int req, int rsp, int missing) { StatsEntry entry=map.get(key); if(entry == null) { entry=new StatsEntry(); map.put(key, entry); } entry.xmit_reqs+=req; entry.xmit_rsps+=rsp; entry.missing_msgs_rcvd+=missing; } Here, different threads might write to different Entries, but since this is only for stats and only on initial startup, this is fine. We'll just lose a few data points... -- 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