Project

General

Profile

Feature #6640

Add tmp area in glideins which is guaranteed to be on local storage in glideins

Added by Igor Sfiligoi almost 6 years ago. Updated over 5 years ago.

Status:
Closed
Priority:
High
Assignee:
Igor Sfiligoi
Category:
Glidein
Target version:
Start date:
07/18/2014
Due date:
% Done:

0%

Estimated time:
Stakeholders:

CMS

Duration:

Description

Some OS mechanisms require local storage to operate,
but glideins may start on a shared file system,
so using that area may not appropriate.
(an example are named sockets)

Let's add a per-glidein tmp subdir in /tmp,
with guaranteed cleanup at glidein termination
(unless hard killed, of course)


Related issues

Blocks GlideinWMS - Feature #6626: Add shared_port_daemon support in the glideinsClosed07/15/2014

History

#1 Updated by Igor Sfiligoi almost 6 years ago

  • Status changed from New to Feedback
  • Assignee changed from Igor Sfiligoi to Parag Mhashilkar

Implemented the feature, now in branch v3/6640.

The directory location is advertised as GLIDEIN_LOCAL_TMP_DIR.

Please review.

#2 Updated by Parag Mhashilkar almost 6 years ago

  • Status changed from Feedback to Resolved
  • Assignee changed from Parag Mhashilkar to Igor Sfiligoi

reviewed, gave feedback, merged changes to branch_v3_2 and master

#3 Updated by Parag Mhashilkar almost 6 years ago

  • Status changed from Resolved to Assigned

I am re opening this ticket because of the issue I notice while testing something else. I see following error

<?xml version="1.0"?>
<OSGTestResult id="glidein_startup.sh" version="4.3.1">
  <operatingenvironment>
    <env name="client_name">Frontend-master-v1_0.main</env>
    <env name="client_group">main</env>
    <env name="user">glidein_pilot</env>
    <env name="arch">x86_64</env>
    <env name="os">Scientific Linux Fermi release 5.8 (Lederman)</env>
    <env name="hostname">fermicloud150.fnal.gov</env> 
    <env name="cwd">/home/glidein_pilot</env>
  </operatingenvironment>
  <test>
    <tStart>2014-07-31T10:47:07-05:00</tStart>
    <tEnd>2014-07-31T10:47:11-05:00</tEnd>
  </test>
  <result>
    <status>ERROR</status>
    <metric name="TestID" ts="2014-07-31T10:47:11-05:00" uri="local">condor_startup.sh</metric>    <metric name="failure" ts="2014-07-31T10:47:11-05:00" uri="local">Config</metric>    <metric name="MissingAttribute" ts="2014-07-31T10:47:11-05:00" uri="local">GLIDEIN_LOCAL_TMP_DIR</metric>
  </result>
  <detail>
     Validation failed in condor_startup.sh.

    Cannot extract GLIDEIN_LOCAL_TMP_DIR from 'glidein_config'
  </detail>
</OSGTestResult>

And looking GLIDEIN_LOCAL_TMP_DIR configuration we force it as required

mac-117966:glideinWMS parag$ grep LOCAL creation/web_base/condor_vars.lst
GLIDEIN_LOCAL_TMP_DIR   C       -               +                               Y       N       @

This will surely break all the existing factories and make the release backwards incompatible. I propose we rethink if this attribute is really required and if not make it optional

#4 Updated by Parag Mhashilkar almost 6 years ago

Just realized the changes are in glidein_startup.sh so I have to upgrade the factory. A simple reconfig wont do. My comment still stands, do we really want to define this attribute in condor_vars as required?

#5 Updated by Igor Sfiligoi almost 6 years ago

  • Status changed from Assigned to Feedback
  • Assignee changed from Igor Sfiligoi to Parag Mhashilkar

Yes, it should be required.

It is "hard coded" in glidein_startup.sh.
If it is not in the config file by the time condor is started, something went terribly wrong.

#6 Updated by Parag Mhashilkar almost 6 years ago

  • Status changed from Feedback to Resolved
  • Assignee changed from Parag Mhashilkar to Igor Sfiligoi

ok i will resolve the ticket

#7 Updated by Parag Mhashilkar over 5 years ago

  • Status changed from Resolved to Closed


Also available in: Atom PDF