Project

General

Profile

Feature #24306

Evaluate non exec-ing singularity in the wrapper to be able to examine the execution and report better errors

Added by Marco Mambelli 4 months ago. Updated 3 months ago.

Status:
New
Priority:
Normal
Assignee:
-
Category:
Glidein
Target version:
Start date:
04/12/2020
Due date:
% Done:

0%

Estimated time:
Stakeholders:
Duration:

Description

The current OSG wrapper is not exec-ing singularity to continue after and inspect if it ran correctly or not.
This may be causing problems w/ signals (sig-term/quit) propagation

This should be evaluated because the additional debugging information would be good for troubleshooting purposes.
This is done touching the file .singularity.startup-ok
Breadcrumbs could be added to provide more information


    # signal our parent that we got here
    touch .singularity.startup-ok

...

        EC=$?
        if [ $EC -ne 0 ]; then
            # was it a Singularity issue or a user job issue?
            if [ ! -e .singularity.startup-ok ]; then
                shutdown_glidein "Singularity encountered an error starting the container" 
            fi
        fi
        if [ "x$GWMS_DEBUG" = "x" ]; then
            rm -f .singularity.startup-ok
        fi
        exit $EC

Evaluate also the use of trace callback

if [ ! -e .trace-callback ]; then
    (wget -nv -O .trace-callback http://osg-vo.isi.edu/osg/agent/trace-callback && chmod 755 .trace-callback) >/dev/null 2>&1 || /bin/true
fi
./.trace-callback start >/dev/null 2>&1 || /bin/true

History

#1 Updated by Marco Mambelli 4 months ago

  • Description updated (diff)

#2 Updated by Marco Mambelli 3 months ago

  • Target version changed from v3_6_3 to v3_6_4


Also available in: Atom PDF