Re: [HTCondor-users] HTCondor-CE Upgrade problems from 3.4.3 to 5.1.3

[Jaime Frey <jfrey@xxxxxxxxxxx>]

Based on this error, the CEâs condor_schedd closed the connection abruptly at the end of authentication/authorization. /var/log/condor-ce/SchedLog should have some data about why it didnât like the submission attempt.

What version of HTCondor is installed on the CE and the EGI monitoring system?

 - Jaime

On Nov 15, 2022, at 10:15 AM, David Cohen <cdavid@xxxxxxxxxxxxxxxxxxxxxx> wrote:

Adding some information:

I'm using the default job router that was installed with HTCondor-CE 3.4.3.

The command failing is part of the EGI monitoring system:

https://argo-mon2.egi.eu/nagios/cgi-bin/extinfo.cgi?type=2&host=tau-htc.hep.tau.ac.il&service=ch.cern.HTCondorCE-JobState-ops

CRITICAL - Job submit result undefined, failed to parse the job ID
1 condor_submit --spool --name tau-htc.hep.tau.ac.il --pool tau-htc.hep.tau.ac.il:9619 /var/lib/gridprobes/ops/scondor/tau-htc.hep.tau.ac.il/gridjob.jdl
Submitting job(s)
ERROR: Failed to connect to queue manager tau-htc.hep.tau.ac.il
SECMAN:2007:Failed to received post-auth ClassAd

JDL([('universe', 'vanilla'), ('executable', 'hostname'), ('transfer_executable', 'true'), ('output', '/var/lib/gridprobes/ops/scondor/tau-htc.hep.tau.ac.il/out/gridjob.out'), ('error', '/var/lib/gridprobes/ops/scondor/tau-htc.hep.tau.ac.il/out/gridjob.err'), ('log', '/var/lib/gridprobes/ops/scondor/tau-htc.hep.tau.ac.il/out/gridjob.log'), ('log_xml', 'true'), ('should_transfer_files', 'YES'), ('when_to_transfer_output', 'ON_EXIT'), ('use_x509userproxy', 'true')])

On Tue, Nov 15, 2022 at 1:50 PM David Cohen <cdavid@xxxxxxxxxxxxxxxxxxxxxx> wrote:

Hi all,

I tried a rolling upgrade from 8.8 to 9.0

We have PASSWORD authentication and upgrading the startd and sched went smoothly.

I just had to comment "# use security:recommended_v9_0"

The problems started when upgrading the HTCondor-CE with:

SECMAN:2007:Failed to received post-auth ClassAd

I continued and upgraded the central manager to make sure it is not a version mismatch, but the problem persists.

Help will be much appreciated,

David