#1 Thread deadlocks

[Anonymous]

I run in an asynchronous environment. Currently
whenever I try to implement swarmcache as my caching
mechanism, I seem to get thread deadlocks.

[André Schild]

Logged In: YES
user_id=119137

at java.lang.Object.wait(Native Method)
- waiting on <0x15f0c3a8> (a java.lang.Object)
at java.lang.Object.wait(Object.java:429)
at org.javagroups.JChannel.connect(JChannel.java:266)
- locked <0x15f0c3a8> (a java.lang.Object)
- locked <0x15f0c3b0> (a org.javagroups.JChannel)
at
org.javagroups.blocks.NotificationBus.start(NotificationBus.java:126)

at
net.sf.swarmcache.JavaGroupsCommunicator.<init>(JavaGroupsCommunicato
r.java:39)
at
net.sf.swarmcache.CacheFactory.<init>(CacheFactory.java:65)
at
com.aarboard.aarcat2.utils.SiteSettings.<init>(SiteSettings.java:202)

[André Schild]

Logged In: YES
user_id=119137

We too see that problem at random.
We have 3 sites who use swarmcache. And it seems as
if they deadlock 1-2 times a week. (The Tomcats
are restarted each night)

See the stackdump pasted above for such a hanging thread.

[André Schild]

Logged In: YES
user_id=119137

We are currently working on this issue.

https://sourceforge.net/tracker/?func=detail&atid=106081&aid=833309&group_id=6081

Andr

[André Schild]

Logged In: YES
user_id=119137

Is this still an issue for you ?

One important thing is, that the host name as seen by JAVA
matches the one the other cache members see via DNS (or
whatever name resolution you use)

In our case, one system was named cs01 and the other one cs02.
But in the reverse DNS for the systems we had the names
cluster3 and cluster4.

So they did never find the other one and the cache startup
failed.

Andr