Re: [Condor-users] MasterLog and ShadowLog errors

[Ian Stokes-Rees <ijstokes@xxxxxxxxxxxxxxxxxxx>]

Hi Zach,

I think you called me this afternoon.  If so, thanks.  For the moment, 
Condor is running.  It has been in this state three times in the past 
few months, but inevitably within a few days or at most a week or two it 
will stop working, and I swear I don't know what changes are made to 
have it all fall apart.  Yesterday I reinstalled Condor (twice) and 
stripped my configuration files (BTW, a 2500 line config file is really 
a pain to work with).  I deleted the log, execute, and spool directories 
to give me (as much as possible) a "starting from scratch" state.  Let's 
see what happens now.

I have a few higher priority tasks on my plate at the moment, but my 
next endeavour with Condor is to get a Mac OS X worker node tied in 
(challenge is that it is on a different sub-net, no NFS mounting, no 
shared users), and then after that we want to have flocking and cross 
submission to our other Condor pool.

The security mechanisms in Condor still perplex me, but I can happily 
report that I've gotten rid of all the "ALLOW = *" entries which, of 
course, I was really unhappy about.  I would much rather have all the 
machines authenticate with X.509 host or service certs, which I get the 
impression is possible, but I haven't had time to look into how to do this.

Cheers,

Ian

--
Ian Stokes-Rees, Research Associate
SBGrid, Harvard Medical School
http://sbgrid.org