[jcifs] 401 and hanging threads after TransportException

[Michael B Allen]

On Mon, 05 Nov 2007 08:26:56 +0100
Ronald Müller <ronald.mueller at infopark.de> wrote:

> Hello,
> 
> I have a problem with jcifs 1.2.17 and up till now I couldn't find a 
> solution in your list. I also tried to use jcifs 1.1.9 and jcifs 1.2.11 
> but this didn't make any difference.
> 
> Sometimes my client/application gets a 
> jcifs.smb.SmbException/jcifs.util.transport.TransportException
> 
> jcifs.smb.SmbException:
> jcifs.util.transport.TransportException: Transport29 timedout waiting 
> for response to 
> SmbComSessionSetupAndX[command=SMB_COM_SESSION_SETUP_ANDX,received=false,errorCode=0,flags=0x0018,flags2=0xC003,signSeq=0,tid=0,pid=50963,uid=0,mid=55,wordCount=13,byteCount=89,andxCommand=0x75,andxOffset=150,snd_buf_size=16644,maxMpxCount=10,VC_NUMBER=1,sessionKey=0,passwordLength=24,unicodePasswordLength=24,capabilities=4180,accountName=g52479,primaryDomain=,NATIVE_OS=Linux,NATIVE_LANMAN=jCIFS]
> 	at jcifs.util.transport.Transport.sendrecv(Transport.java:76)
> 	at jcifs.smb.SmbTransport.send(SmbTransport.java:606)
> 	at jcifs.smb.SmbSession.sessionSetup(SmbSession.java:277)
> 	at jcifs.smb.SmbSession.send(SmbSession.java:233)
> 	at jcifs.smb.SmbTree.treeConnect(SmbTree.java:154)
> 	at jcifs.smb.SmbSession.logon(SmbSession.java:169)
> 	at jcifs.smb.SmbSession.logon(SmbSession.java:162)
> 	at jcifs.http.NtlmHttpFilter.negotiate(NtlmHttpFilter.java:182)
> 	at jcifs.http.NtlmHttpFilter.doFilter(NtlmHttpFilter.java:114)
>          ...
>          ...
>          ...
> 
> First of all I've no idea why the client sometimes gets this exception. 
> In my logfile I can see that shortly before and shortly after this 
> exception my program can authenticate users at the domain controller 
> without any problems. And after getting this error it works for hours 
> without any problems. So why at just that moment it gets this exception? 
> Neither the domain controller nor the server my application is running 
> on have a high load at this time. Do you have an advice what to check?

Try increasing the timeout to something very large to eliminate the possiblity that it's a simple network fault.

> The second problem around this exception is the following behaviour: If 
> one thread runs into this exception also all other threads - trying to 
> get an authentication at the domain controller - hang until the first 
> thread reaches the timeout (10 seconds). Ok, as far I learned within the 
> last days reading your list this behaviour could be caused because jcifs 
> is synchronizing all authentication requests so that if one hangs also 
> all other requests hang (waiting for the timeout of the first request). 
> Is my assumption correct?

If you're using one specific domain controller then yes. Sounds like the correct behavior to me.

> But more irritating: All of these hanging/waiting threads get an 401 the 
> users are seeing in their browsers and they are asked (via basic auth 
> promt) to enter login/password. Why even the other threads (not getting 
> the exception) get an 401? Can it be that all these threads/requests use 
> the same socket and after the first request gets the exception also the 
> socket dies? Is there a way to force this threads to use another socket 
> so they don't get an 401?

Not sure about the 401 (don't recall well enough how the filter works).

Never heard of a basic auth prompt popping up but if IE cannot authenticate successfully it will pop up the "Network Password" dialog.

Mike

-- 
Michael B Allen
PHP Active Directory SPNEGO SSO
http://www.ioplex.com/