Strange behaviour with perlbal

Oezcan Acar acar at
Fri May 18 07:39:51 UTC 2007


I installed the latest perlbal from svn. We use the following configuration:

1. Request arrives at Cisco Pix
1.1 per static command of pix os the official IP address ist mapped to internal lan address
2. Pix delivers the request to alteon d3 loadbalancer (to virtual address of perlbal cluster)
3. Alteon forwards the request to one of the perbal instances in LAN.
4. Perlbal is configured as reproxy. The request ist forwarded to resin application server.

If you check the webapp via browser and refresh it couple times each third or fourth request hangs. First, second or third request is responded very quickly but fourth or fifth request hangs and you get delayed response.

This behaviour is not seen if we use perlbal without alteon, directly over pix. We use alteon for our webcluster and such a bahaviour not seen ever.

Do someone have a hint or experience of such a configuration?


Mit freundlichen Grüßen


Özcan Acar, Dipl. Inf(FH)
Smart Web Business AG

Im Niederfeld 8
65462 Gustavsburg

+49 171 57 33 66 5
acar at

More information about the perlbal mailing list