[omniORB] bug in bulk messages processing?

[Duncan Grisby]

On Sunday 15 March, Serguei Kolos wrote:

> In my configuration I have a single server application serving many
> concurrent multi-threading clients which are sending requests to this
> server from several threads concurrently. I set oneCallPerConnection
> parameter to 0 on the clients. In this configuration very often I have
> some clients timing out while the server is not really under heavy
> load. To investigate the situation I have run the server and the
> clients with the traceLevel 35 and I have found a possible indication
> that when a bulk of messages arrives over the same connection to the
> server and the server splits them into separate buffers then it looks
> like only first message gets processed and others get lost.

Well, the support for splitting messages into separate buffers
definitely works, with all calls being processed correctly, both for a
simple example I just tried and for a much more complex application I've
looked at traces for. So, the problem is either something specific to
the circumstances you're in, or something completely different.

It would be useful if you could send me a complete trace from
-ORBtraceLevel 30 -ORBtraceInvocations 1 -ORBtraceInvocationReturns 1
-ORBtraceThreadId 1 -ORBtraceTime 1 . That will give a much better idea
of what's going on.

Cheers,

Duncan.

-- 
 -- Duncan Grisby         --
  -- duncan at grisby.org     --
   -- http://www.grisby.org --