[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]

Re: [HTCondor-users] Problems with submit node in version 8.6.0



On 2/1/2017 9:13 AM, Feldt, Andrew N. wrote:
Zach,

Update - I had neglected to try the ‘-allusers’ flag - with this flag I
find that I can run condor_q from a non-submit node without any problem
(and without the fix you suggest) in our environment.  I.e. ‘condor_q
-allusers -g’ works just fine for us.

Andy

So if you want, you can have condor_q do the "-allusers" flag behavior by default by placing the following into your condor_config file:

  CONDOR_Q_ONLY_MY_JOBS = False

regards,
Todd




On Feb 1, 2017, at 9:04 AM, Feldt, Andrew N. <afeldt@xxxxxx
<mailto:afeldt@xxxxxx>> wrote:

	This sender failed our fraud detection checks and may not be who they appear to be. Learn about spoofing <http://aka.ms/LearnAboutSpoofing> 	Feedback <http://aka.ms/SafetyTipsFeedback>

Zach,

Unfortunately, this does not work in all environments.  We are an
NFS/NIS set of systems.  Without the fix you have suggested, condor_q
works for the submitter or a queue super user directly from the
submitting host but not from any other node (where one would have to
use the -g flag).  However, after setting the two variables below,
condor_q does not even work from the submitting node.

An analysis of the logs (without the fix) shows that ‘condor_q -g’
from a non-submitting node fails because no authentication method
works.  We use ‘FS’ authentication which works for 8.4 in our NFS
environment on RHEL 7.3 with SELinux enabled, but fails for 8.6.0.

Andy

*Andy Feldt*
/Senior System Support Programmer/
/Affiliate Assistant Professor/
Homer L. Dodge Department of Physics & Astronomy
The University of Oklahoma

On Feb 1, 2017, at 6:27 AM, Zach Miller <zmiller@xxxxxxxxxxx
<mailto:zmiller@xxxxxxxxxxx>> wrote:

This is indeed a problem, related to the new behavior of having
condor_q show only the jobs for the user running condor_q.

Unfortunately, it’s not as simple as running ‘condor_q –allusers’ to
return to the old behavior.  For the time being, if you want to use
8.6.0 you’ll need to change:

      SEC_DEFAULT_AUTHENTICATION = OPTIONAL
      SEC_DEFAULT_NEGOTIATION = OPTIONAL

This does add some overhead in terms of network traffic, but it does
work.  We’ll take a deeper look at how to address this, but in the
meantime I just wanted to put that workaround out there.


Cheers,
-zach


On 2/1/17, 12:22 AM, "HTCondor-users on behalf of
Greg.Hitchen@xxxxxxxx <mailto:Greg.Hitchen@xxxxxxxx>"
<htcondor-users-bounces@xxxxxxxxxxx
<mailto:htcondor-users-bounces@xxxxxxxxxxx> on behalf of
Greg.Hitchen@xxxxxxxx <mailto:Greg.Hitchen@xxxxxxxx>> wrote:

   Hi All

   Thought I'd have a look at the latest stable release but have had
some issues/problems.
   I have tried this on win7 32 bit with 32 bit condor 8.6.0, win7 64
bit with 32 bit condor 8.6.0,
   win2008 server 64 bit with 32 bit condor 8.6.0, and win2008 server
64 bit with 64 bit condor 8.6.0
   Lastly I have tried it on Ubuntu1604  64 bit.
   I have the same problem with all of the above combinations.

   Security: host based

   I'm testing the 8.6.0 version using the same condor_config files
that work for 8.4.4 (and previous versions).

   If I try a condor_q command on the submit node  for 8.6.0 I get:

condor_q
   -- Failed to fetch ads from:
<152.83.64.39:9618?addrs=152.83.64.39-9618&noUDP&sock=4488_72cc_3> :
WIN2008-GJH9-VC.nexus.csiro.au <http://win2008-gjh9-vc.nexus.csiro.au/>

   If I run the 8.4.4 condor_q executable it works and I get:

c:\condor-8.4.4\bin\condor_q
   -- Schedd: WIN2008-GJH9-VC.nexus.csiro.au
<http://win2008-gjh9-vc.nexus.csiro.au/> : <152.83.64.39:9618?...
    ID      OWNER            SUBMITTED     RUN_TIME ST PRI SIZE CMD

   0 jobs; 0 completed, 0 removed, 0 idle, 0 running, 0 held, 0 suspended

   In fact I can run the 8.4.4 version of condor_q from any other
submit node using the -name option
   and that works OK as well.

   Turning on FULL_DEBUG gives the following in the schedd log file:
   01/31/17 15:27:33 Calling Handler
<DaemonCommandProtocol::WaitForSocketData> (6)
   01/31/17 15:27:33 Failed to read end of message from
<152.83.64.39:60746>; 310 untouched bytes.
   01/31/17 15:27:33 AUTHENTICATE: handshake failed!
   01/31/17 15:27:33 DC_AUTHENTICATE: Our security policy is invalid!
   01/31/17 15:27:33 Return from Handler
<DaemonCommandProtocol::WaitForSocketData> 0.000196s

   The above bit about "Failed to read end of message" seems strange?
As does the AUTHENTICATE
   lines. For reference we have the following config lines some of
which should turn off authentication?
   QUEUE_ALL_USERS_TRUSTED = True
   SEC_DEFAULT_AUTHENTICATION = NEVER
   SEC_DEFAULT_AUTHENTICATION_METHODS = CLAIMTOBE
   SEC_DEFAULT_AUTHENTICATION_TIMEOUT = 20
   SEC_DEFAULT_NEGOTIATION = NEVER

   If I install 8.6.0 from the msi windows installer file (rather
than just unzipping), and therefore use
   the generated condor_config file then condor_q works OK.

   I have painstakingly compared the output from condor_config_val
-dump- verbose for 8.6.0
   using both our 8.4.4 config files and the 8.6.0 generated config
files and cannot see many
   differences, and those don't "appear" to be important.

   I've tried using the generated condor_config and renaming our
8.4.4 condor_config to condor_config.local
   but then the condor_q error is back again.

   Any help/suggestions greatly appreciated.

   Cheers

   Greg

   _______________________________________________
   HTCondor-users mailing list
   To unsubscribe, send a message to
htcondor-users-request@xxxxxxxxxxx
<mailto:htcondor-users-request@xxxxxxxxxxx> with a
   subject: Unsubscribe
   You can also unsubscribe by visiting
   https://lists.cs.wisc.edu/mailman/listinfo/htcondor-users

   The archives can be found at:
   https://lists.cs.wisc.edu/archive/htcondor-users/



_______________________________________________
HTCondor-users mailing list
To unsubscribe, send a message to htcondor-users-request@xxxxxxxxxxx
<mailto:htcondor-users-request@xxxxxxxxxxx> with a
subject: Unsubscribe
You can also unsubscribe by visiting
https://lists.cs.wisc.edu/mailman/listinfo/htcondor-users

The archives can be found at:
https://lists.cs.wisc.edu/archive/htcondor-users/

_______________________________________________
HTCondor-users mailing list
To unsubscribe, send a message to htcondor-users-request@xxxxxxxxxxx
<mailto:htcondor-users-request@xxxxxxxxxxx> with a
subject: Unsubscribe
You can also unsubscribe by visiting
https://lists.cs.wisc.edu/mailman/listinfo/htcondor-users

The archives can be found at:
https://lists.cs.wisc.edu/archive/htcondor-users/



_______________________________________________
HTCondor-users mailing list
To unsubscribe, send a message to htcondor-users-request@xxxxxxxxxxx with a
subject: Unsubscribe
You can also unsubscribe by visiting
https://lists.cs.wisc.edu/mailman/listinfo/htcondor-users

The archives can be found at:
https://lists.cs.wisc.edu/archive/htcondor-users/



--
Todd Tannenbaum <tannenba@xxxxxxxxxxx> University of Wisconsin-Madison
Center for High Throughput Computing   Department of Computer Sciences
HTCondor Technical Lead                1210 W. Dayton St. Rm #4257
Phone: (608) 263-7132                  Madison, WI 53706-1685