SCIENTIFIC-LINUX-USERS Archives

June 2014

SCIENTIFIC-LINUX-USERS@LISTSERV.FNAL.GOV

Options: Use Monospaced Font
Show Text Part by Default
Show All Mail Headers

Message: [<< First] [< Prev] [Next >] [Last >>]
Topic: [<< First] [< Prev] [Next >] [Last >>]
Author: [<< First] [< Prev] [Next >] [Last >>]

Print Reply
Subject:
From:
Joseph Areeda <[log in to unmask]>
Reply To:
Joseph Areeda <[log in to unmask]>
Date:
Tue, 24 Jun 2014 09:28:58 -0700
Content-Type:
text/plain
Parts/Attachments:
text/plain (52 lines)
On 06/24/2014 06:56 AM, Larry Linder wrote:
> On Tuesday 24 June 2014 8:54 am, Joseph Areeda wrote:
>> Thanks Stephan,
>>
>> On 06/24/2014 01:07 AM, Stephan Wiesand wrote:
>>> On 2014-06-24, at 6:48, Joseph Areeda <[log in to unmask]> wrote:
>>>> I have a C++ program that runs on multiple systems.  It uses a
>>>> proprietary network protocol contained in a shared object.
>>>>
>>>> On one of the systems I get this error regularly but not often enough to
>>>> use a debugger:
>>>>
>>>> NDS library error: Resource temporarily unavailable
>>>>
>>>> It only seems to happen on one system, my workstation.  I've reinstalled
>>>> the library.  I have Googled my heart out and while I see the error
>>>> reported in other packages I haven't found anything that explains what
>>>> it means.  NDS is the name of the service (Network Data Service).
>>>>
>>>> The only hints I've gotten suggest it might mean the network interface
>>>> itself might be involved but nothing else seems to have a problem.  If
>>>> it were a bug in the library I'd expect to see it on the other systems
>>>> which are in production.
>>>>
>>>> Any clues as to what it means or where to read up on it would be greatly
>>>> appreciated.
>>> Some library call returned EAGAIN. The prime suspect is usually fork(2),
>>> but in the case of a network library, I'd look at send(2) first.
>>>
>>> Hth
>>> 	Stephan
> Do you have a managed switch behind a router in system?
> Maybe a 1G Hz router feeding a 100K Hz router?
> Most newer boxes have a 1GHz NIC (built in) in them.   We have a managed 1 G
> Hz switch that is managed and we have a Motorola router that is less than 1
> GHz with factory default set up.   The switch shows up on our network as a
> device!   I sort of wondered about it as a cause of our stack up (slowness)
> sometimes in the afternoon.   I have never seen much discussion about a
> managed switch and network performance.
>
> Larry Linder
> Larry Linder
Well thanks again Stephan, my problem was indeed a socket timeout 
problem.  This project is a proxy server for the proprietary protocol, a 
stand alone threaded java to C++ interface.  Some moron (me) put the 
timeout on listen connection instead of the client session socket.  D'Oh!

Larry,  I do not use a managed switch.

Best,
Joe

ATOM RSS1 RSS2