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

Re: [Condor-users] can't get Condor-G job to run



On Jun 24, 2005, at 6:21 AM, Dr Ian C. Smith wrote:

On Jun 9, 2005, at 4:35 AM, Dr Ian C. Smith wrote:


--On 08 June 2005 10:36 -0500 Jaime Frey <jfrey@xxxxxxxxxxx> wrote:


[snip]



Can you try starting the gahp_server at the command line and typing the
following line at it:
CACHE_PROXY_FROM_FILE 1 /tmp/x509up_u41269



That is the command that's failing. What does the gahp server reply? If
the command succeeds, it will print a single 'S'.




No I just get a single 'E' printed.



Hmm, that indicates a syntax error. Make sure there are no spaces at the
beginning or end of the line and only a single space between the strings.



Ooops my mistake. I missed the "1" out. I do get a single 'S' printed.



OK, so that means there's something different between the environment of 

your command line and that of Condor >>(environment here meaning the entire execution environment, not just the env variables). The usual culprit is 

Condor not finding the same CA certificates directory, though I've never 

seen the specific error message you're >getting.


See what Globus/X509-related environment variables are set in your shell 

and relaunch the condor_master with >>>them set the same.

Apologies for the break in getting back to this. I tried pulling out
my Globus env vars and exporting them as root before starting condor_master
on the submit host. Unfortunately this has brought down the whole pool
(see other post). Assuming that I can ever get it working again ...

Where does Condor get the Globus config from ? In my .profile I have


export GLOBUS_LOCATION=/apps/globus
. $GLOBUS_LOCATION/etc/globus-user-env.sh


but I can't see anything like this in the condor_config file ? Does
it need to be in the shell startup script for whatever account runs
the condor daemons ? Condor-G works fine on the central manager
and this has pretty much the same setup.

The only Globus-relate external dependency for Condor-G should be the trusted CA certificates directory. The GSI code looks for it in the usual way, checking each of these locations in order:
* $(X509_CERT_DIR)
* /etc/grid-security/certificates
* $(GLOBUS_LOCATION)/share/certificates
* $(HOME)/.globus/certificates (this path may be slightly wrong, and it may belong further up the list)

You can tell Condor to use a specific directory with the GSI_DAEMON_TRUSTED_CA_DIR config file parameter, like so:

GSI_DAEMON_TRUSTED_CA_DIR = /apps/globus/share/certificates

+----------------------------------+---------------------------------+

|            Jaime Frey            |  Public Split on Whether        |

|        jfrey@xxxxxxxxxxx         |  Bush Is a Divider              |

|  http://www.cs.wisc.edu/~jfrey/  |         -- CNN Scrolling Banner |

+----------------------------------+---------------------------------+