Groupwise migr.: hang when using more than one thread

[adrighem]

Hi,
We're trying to migrate from groupwise to zimbra. During our test migration, the groupwise migration tool hangs when using more than one thread. This also happens when using one thread on more than one migration client (multiple PCs). It happens sooner when using more threads, so it looks like it has something to do with the number of connections.

Our install is zimbra on Redhat 5, two server setup, auth to AD.

I also tried putting a reverse proxy between the migration PCs and the zimbra mailbox server, as I was seeing a lot of sockets in CLOSE_WAIT on the windows XP PCs.

This is a real pain. If we have to migrate one mailbox at a time, our migration window will grow beyond our maximum of 48 hours (one weekend).

[adrighem]

Info: the load on the server is not high. Between 0.4 and 1.6.
Firewalls are all turned off.

Turned off ipv6 -> no change
Tried virtual and physical windows machines -> no change
Turned off convertd -> no change
Turned off redolog -> no change
Turned up jetty connection timeout on admin interface to 300 seconds for normal and 100 seconds during load. -> no change

The mailbox.log just stops one connection at a time. At first all ten (for example) connections will be in the logs.. One by one they will die offf. Only one will continue....sometimes...

[adrighem]

Attached mailbox log