Got "Exhausted available memory" with Orbix/iona 3.0.1

[jaakovb]

Hi,

is anybody know the circumstances that cause a process to fail with the
message "Exhausted available memory" ?

I'm working on Unix/Solaris 2.6 with Iona/Orbix 3.0.1.

I don't know if at the same time the available memory in the system was
enough to run the application.

Thanks for any tips.

Yaakov Berkovitch

jaa...@orckit.com

[Daniel Garrido]

Hello,

I had the same problem using callbacks. The problem could be originated by
different speeds in client and server (client slower). Solutions?

jaakovb escribió:

[Barry Haddow]

Hi

You can get this type of error if Orbix receives a corrupted message, or
receives an IIOP message when it is expecting a POOP (Orbix protocol)
message. The latter problem can occur with the daemon under some
circumstances - look for the article in the Iona KB. Some possible
sources of corrupted messages are incorrectly coded transformers,
trying to use SSL-IIOP with an application that only understands
plain IIOP or piggy-backing data incorrectly.
It is unlikely that you are genuinely out of memory - It is more likely
that
Orbix is confused by the message corruption and trying to allocate
a ridiculously large buffer.

hope that helps
Barry Haddow
Orbism
http://www.orbism.com

[pant...@my-deja.com]

In article <873uff$i6t$1...@news.netvision.net.il>,

"jaakovb" <jaa...@orckit.com> wrote:
> Hi,
>
> is anybody know the circumstances that cause a process to fail with
the
> message "Exhausted available memory" ?
>
> I'm working on Unix/Solaris 2.6 with Iona/Orbix 3.0.1.
>
> I don't know if at the same time the available memory in the system
was
> enough to run the application.
>
> Thanks for any tips.
>
> Yaakov Berkovitch
>
> jaa...@orckit.com
>
>

I'm also working under Unix/Solaris 2.6 with
Iona/OTM Orbix 3.0.1. with the client and server
written in C++.

I run my two processes (client & server ) and
Iona's logging in the server reports
'Exhausted available memory'. I too think it's
because the server while trying to marshall the
message and deliver it my server implemenation
asks for a ridiculously large chunk of memory.

Here's 'essentially' what the relevant
idl is like:

getMyNestedStruct( out MyNestedStruct xx );
setMyNestedStruct( in MyNestedStruct xx );

So, the client connects to a server. It
invokes getMyNestedStruct and receives a piece
of memory containing a MyNestedStruct.
Then the client invokes setMyNestedStruct,
'sending' that same piece of memory back to the
server.
The server craps out with 'Exhausted available
memory' as soon as
CORBA::Orbix.continueThreadDispatch()
is called.

Any ideas, solutions, or workarounds appreciated.

Regards,

Peter.

Sent via Deja.com http://www.deja.com/
Before you buy.

[Chris McCauley]

Hi,

In most cases the problem is either the stubs and skeletons are out of
sync so rebuild the project or
orbix is receiving IIOP messages on a channel which it thinks is using
POOP. This often shows
up in the Orbix daemon log as a message size which equates to the
numeric equivalent of the letters
GIOP.

Chris

[pant...@my-deja.com]

The problem described in my original posting has been
resolved. The problem was due to an uninitialized
idl union which preceded a sequence
within a deeply nested structure.
This resulted in improper de-marshalling in
the server side so that the server was reading
a string from the CDR data stream and taking it to
mean the length of a sequence.
Example, when it read 'PQR' ( 0x50,0x51,0x52,0x00 ) it
treated this as a four byte sequence length.

In article <38A00A44...@ilim.com>,