[omniORB] omniORB slow transfer when clientCallTimeOutPeriod != 0

Igor Lautar igor.lautar at hermes-softlab.com
Tue Aug 11 14:07:06 BST 2009


Hi,

On Tuesday 11 August 2009 12:18:09 Sampo Ahokas wrote:
> We thought about the timeout, but it's not really applicable to us.
> There's just no way of knowing how long the call takes.
>
> In our case this problem seemed to only occur when the server died or
> had to be terminated unexpectedly and thus the we decided to put the
> effort simply into making the server as robust as possible.

Yeah, sounds like what we are experiencing.

In our case, its not a problem of server dying, but intentional restart or a 
complete crash of server (power loss situation).

One thing crossed my mind now. Did you experience these hangs only on crash or 
also if server was shutdown properly (ie. ORB was properly de-initialized)?

In our case, due to other reasons (speed needed, and proper de-init takes time 
- do not see a point as we are shutting down anyway), we have decided to not 
shutdown orb before exiting from server program. Maybe if we shutdown the ORB, 
problem would not be reproduced? Have to test this, but thx for idea ;)

Regards,



More information about the omniORB-list mailing list