Re: [HTCondor-users] Condor_kbdd.exe crashes/doesn't start if no network is available

[Michael Pelletier <Michael.V.Pelletier@xxxxxxxxxxxx>]

I do the same thing - all config is under /home/condor, and the local config files just point the daemons in that direction. Thankfully I don't have network problems very often.

You definitely don't need to write your own update service. Mark Burgess already did that nearly 25 years ago. The approach I'd use to maintain a sitewide config on the local disk would be CFEngine - it's fairly easy to set it up to maintain two directories in the same state, and the update would be pulled by the client once the network comes back online - and you'd just need a promise to reconfig the daemons once that happens.

	-Michael Pelletier.

> -----Original Message-----
> From: HTCondor-users [mailto:htcondor-users-bounces@xxxxxxxxxxx] On Behalf
> Of Michael Schwarzfischer
> Sent: Friday, May 19, 2017 7:20 AM
> To: HTCondor-Users Mail List <htcondor-users@xxxxxxxxxxx>
> Subject: Re: [HTCondor-users] Condor_kbdd.exe crashes/doesn't start if no
> network is available
> 
> Hi all,
> 
> Thanks for your ideas and comments! I just found out why everything
> doesn't work without network.
> It turned out that it is quite a stupid reason: In our installation we
> refer to a global config file sitting on a network share.
> Naturally, all the deamons crash without access to that file. I hope I
> didn't miss anything, but may I suggest to include a logging about that
> exit error in the future. I only found out because I managed to build and
> debug my own condor version :)
> 
> Anyhow, is there some workaround for that problem?
> Is there a possibility to cache the global config file locally? How can I
> guarantee that it updates on all clients when I update the global version?
> Do I need to write my own update service?
> 
> Alternatively, we may think about a wrapper which checks for the
> availability of the file before starting kbdd.
> 
> Thanks again,
> Best,
> Michael