Print

Print


  [Post Gregory's response to the list, he replied only to me. See below.]

On Mon, Aug 15, 2005 at 02:26:55PM -0400, Gregory J. Sharp wrote:
> 
> On 15Aug2005, at 14:03 , Peter Elmer wrote:
> > Gregory, are you saying that the connections fail completely?
> 
> In valgrind the connections fail completely. Peter is correct: we are  
> using the XrdPosix interface.
> 
> For reference, when I use valgrind I get the following messages:
> 
> 050815 14:17:02 001 Xrd: Create: (C) 2004 SLAC INFN XrdClient 0.3
> 050815 14:17:05 001 Xrd: ReadPartialAnswer: Error reading msg from  
> connmgr (server [lnx243.lns.cornell.edu:1094]).
> 050815 14:17:05 001 Xrd: ReadPartialAnswer: Calling HandleServerError
> 
>   **** At this point HandleServerError recursively redirects to  
> another data server. Since I only have one data server, it goes back  
> to the same one, which fails again ...
> 
> 050815 14:17:05 12087 Xrd: XrdClientSock::RecvRaw: Error reading from  
> socket: Resource temporarily unavailable
> 050815 14:17:05 12087 Xrd: XrdClientMessage::ReadRaw: Error reading 8  
> bytes.
> 050815 14:17:05 001 Xrd: ReadPartialAnswer: Error reading msg from  
> connmgr (server [lnx243.lns.cornell.edu:1094]).
> 050815 14:17:05 001 Xrd: ReadPartialAnswer: Calling HandleServerError
> 
>   **** Now note below that after starting the poll() call in  
> RecvRaw, somebody called Disconnect() - I can't figure out who did that.
> 
> 050815 14:17:05 12087 Xrd: XrdClientSock::RecvRaw: since we entered  
> RecvRaw, socket file descriptor has changed to -1
> 050815 14:17:05 12087 Xrd: XrdClientMessage::ReadRaw: Error reading 8  
> bytes.
> 
>   **** and this recurses until we hit the redirect retry limit. It  
> tries to connect to the server again, gets more failures trying to  
> read and shortly thereafter the whole thing hangs.
> 
> When I run without valgrind there are no errors (yeah, I know, "so  
> don't do that" :-)
> 
> My suspicion is some sort of timing issue, but if you have no trouble  
> using valgrind when running without XrdPosix, then perhaps the  
> problem lies there.
> 
> --
> Gregory J. Sharp                   email: [log in to unmask]
> Wilson Synchrotron Laboratory      url: http://www.lepp.cornell.edu/ 
> ~gregor
> Cornell University                 ph:  +1 607 255 4882
> Ithaca, NY 14853                   fax: +1 607 255 8062
> 


-------------------------------------------------------------------------
Peter Elmer     E-mail: [log in to unmask]      Phone: +41 (22) 767-4644
Address: CERN Division PPE, Bat. 32 2C-14, CH-1211 Geneva 23, Switzerland
-------------------------------------------------------------------------