(r384) StatusReport < Imager

Tags: view all tags
<font FACE="Arial-BoldMT">
<p ALIGN="center"><font size="5" color="#48454B">Imager-Tech Weekly 
Meeting & Open Problems List</font></p>
<p ALIGN="LEFT"><br>
<b><font color="#48454B">Every Tuesday @ 11:30-12:30 <br>
Core Attendees: Robert Bridson(faculty rep), Glen Lee(tech staff rep), Brad Atcheson(linux/mac rep), Cheryl Lau(meeting coordinator), Ernesto Torres(software manager)<br>
Venue: 8th floor lounge</font></b></p>
<hr>
</font>

<!--

---++++ [Template] Title
---++++++ Date Opened: mmm dd, yyyy
---++++++ Impact: 1 = workaround identified, 2 = inconvenience, 3 = research affected, 4 = research might not be completed, 5 = critical/large scale impact
---++++++ Contact Person: Name
---++++++ Issue: Description (help desk number, if assigned)
---++++++ Updates: 
   * Update 1.

-->


---++++ Imager Disk Space
---++++++ Date Opened: Jan 08, 2012
---++++++ Impact: 3 = research affected
---++++++ Contact Person: Robert
---++++++ Issue: The imager disk usage report from Jan 8, 2012 showed imager at 95%.  imager/local is the top dir, followed by project/AnimationPhysics.
---++++++ Updates: 
   * Jan 10: Robert will check disk usage with df daily this week.
   * Jan 08: (email) Robert is in contact with Anika who has posted a large dataset on the imager webpage which is in imager/local.
   * Jan 08: (email) Robert contacted help to try and figure out what's going on with imager/local.  Robert also asked Hagit to clean up.



---++++ ICICS Cluster Scheduling
---++++++ Date Opened: Nov 08, 2011
---++++++ Impact: 3 = research affected
---++++++ Contact Person: Hagit, Anika
---++++++ Issue: Hagit and Anika are using the ICICS cluster and have had issues with job scheduling.  If a long job is scheduled, it will use the entire cluster, making the cluster unavailable until that job finishes.  This means that short jobs will have to wait until the long job finishes.  Is there a way to change the scheduling policy so that short jobs can still run while long jobs are running?
---++++++ Updates: 
   * Nov 22: Glen will see if Dave is in contact with Hagit.



---++++ TSE
---++++++ Date Opened: Oct 14, 2011
---++++++ Impact: 1 = workaround identified
---++++++ Contact Person: Cheryl
---++++++ Issue: When using Power Point through TSE on a Linux machine, viewing the slide show is not possible and also causes the program to crash (unsaved work is lost).  Allan mentioned that he could not view videos through tse.
---++++++ Updates: 
   * Nov 8: No update. Glen will be documenting the workaround.
   * Oct 25: Not a print spooler issue since all TSE servers exhibit same problem. Probably a bug in TSE display drivers. Other users report similar issues. Workaround identified: first click the book-like icon next to the presentation icon, then click the presentation icon.
   * Oct 18: The !PowerPoint problem is not due to the print spool problem, and Glen will look into this.
   * Oct 18: The tse server crashes often due to problems with the print spool.  The tech staff knows about this problem and is working on it.




---++++ Meeting notes
   * Check [[ConferenceFreezeDates][Conference Freeze Dates]] page monthly and notify tech staff of any new freezes.
   * Check imager disk space list monthly for users over 25GB.
   * Nov 8: stray simulation currently generating large amounts of data on imager.  Robert will ask Hagit to clean up or we may have to resort to quotas.
   * No replacement yet identified for Windows rep.
   * Siggraph freeze: Nov 15-Jan 20.  Update freeze list machines on Sunday before the freeze.
   * Large directory /imager/local appeared on the disk usage email. Contents are 10-20 years old.
Topic revision: r384 - 2012-01-10 - cherylsl