You are viewing an old version of this page. View the current version.

Compare with Current View Page History

« Previous Version 2 Next »

IEPM Tasks

Last update: September 4, 2006,Archive
Awaits something, also provides a start of wait date
Done or Drop is deleted when it is > a month old.
- Person(s) responsible
Task being worked on or to be discussed at group meeting
Changes

Action Items

  1. Terapaths
    1. Netflow (see http://iepmbe.bnl.gov/netflow/index.html) - Yee
      1. Talk to John H to find out his needs - Yee
      2. Try and make work on non-Firefox browsers (DOM needs fixing) - Yee
      3. Add spider and pie charts - Yee
      4. Discuss with Connie how to get permanent exec level plots - Yee, Connie, Les
  2. PingER
    1. Make sure Maxim has all the latest monitoring nodes - Jerrod
    2. Get ping-data.pl working at sfsmds2.vsnl.in - Jerrod
  3. Transport services evaluation - Yee Ting Li
    1. Work with Microsoft - Yee, Les
      1. Get latest privates
      2. Decide what is needed for stage 2
  4. MonALISA (no progress 3/12/06, awaits iepm-bw OWAMP integration, keeping servers running) - Connie
    1. Upload selected data (initially IEPM data from BNL, SLAC, Caltech, CERN) using a single object for efficiency (awaits Iosif's new version of ML/APMon) - Adnan, Iosif
    2. Figure out how to display IEPM monitoring hosts and their data - Fawad, Aziz
    3. Project defined and assigned to Akbar and Waqar (3/11/06) - Akbar, Waqar
  5. IEPM-BW
    1. Work with DESY to get new monitoring host (contacted Kars 7/20/04, Kars going on 2 weeks vacation then Jerrod is away, time to re-start 8/26/04, wait for v3, Jerrod sent email reminder 3/25/05, Kars will be here later this month (27th April '06), Jerrod contact him before he arrives) (awaits V3 of iepm-bw) - Jerrod
    2. Make FZK an IEPM Monitoring node - Connie
      1. Get contact for Connie (sent email 8/22/06, now awaits Connie) - Les
    3. Update metrics used
      1. ID and add more targets for pathload - Connie, Jerrod
    4. Get distribution kit for iepm monitoring nodes to install & configure - Jerrod
      1. Update pre-reqs document - Jerrod

      2. Build pacman procedure so admin can do own install (now works to make the database, next step is to create the tables, and copy over and configure the crontabs 3/23/06) [dropped 8/21/06] - Jerrod

      3. After re-think divide task up between what pacman does well, and script the rest [Dropped pacman 8/21/06]

        1. Develop on Taiwan (start 4/17/06) - Jerrod
    5. Write script to use ssh to get the configurations of IEPM monitor and remote hosts (in progress 4/26/06, will revisit Sep '06) - Jerrod
    6. Get architecture of remote nodes and create a web page (wil get back to in Sept 06) - Jerrod
    7. Do we want to get reverse traceroutes (at least where we have reverse traceroute servers, awaits time) - Connie
    8. Compare pathchirp and pathload - Connie
      1. Make up a proposal (see if we need it) - Connie, Adnan
    9. Bugs
      1. Fix up TCP receive buffer sizes, add sanity checks (in progress 4/26/06, Connie will talk to Yee to understand 8/22/06) - Connie
    10. Traceanal - Yee, Asif
      1. Integrate new topology into web server - Yee
        1. Identify the most used routes - Asif
        2. Integrate with pathneck to color links based on speed - Asif
        3. Rendering of topology much slower on www.slac.stanford.edu - Yee
      2. Prepare distributable version of traceanal - Yee
    11. Alert
      1. Look at multivariate event detection (collect data for SLAC, BNL, Caltech pathchirp, thrulay,ping) - Adnan
        1. Need to extend pathload to other sites - Connie
        2. Run plateau on the data for min-RTT, thrulay, pathchirp - Mahesh
        3. Apply to PCA to the same data
      2. Look at improvements to plateau
        1. Ability to find step ups - Adnan
        2. Extend to allow up & down then compare down with original - Adnan
        3. Allow for small number of samples (e.g. at start) - Mahesh
      3. Look at other detection algorithms and compare
        1. Holt-Winters - Les, Mahesh, Felipe
          1. Go back 7 weeks - Mahesh
          2. Check unusual results - Mahesh
          3. Consider other ways to optimize parameters - Mahesh
        2. Neural networks
        3. KS
          1. Look at making points before larger than points after- Akbar
      4. Prepare table of canonical events and how various algorithms react - Adnan
        1. Build case studies of email events (how is this coming on?)- Adnan
      5. Look into host monitoring/isolation
        1. Look at installing LISA/APMon at monitoring sites so can eliminate events caused by host congestion
        2. Ganglia
        3. Nagios
        4. Monitor NIC errors
      6. Look at how to use PerfSONAR - Adnan
      7. Look at detecting outages for ping - Connie
        1. Analyze what constitutes a significant outage - Connie
      8. Understand cause of delayed alerts and see if can improve - Connie
      9. Diagnose events - Adnan
      10. Extend database to add trigger start date/time, trigger detection date/time in database - Connie
  6. Install WANMON as IEPM web server - Yee

    1. Port CGI-WRAP [Done 8/10/06]- Les, Yee

    2. Get NFS and AFS accesss - Yee
    3. Get approval for externally visible web server
    4. Get& traceroute.pl and pingtable.pl working and in production
  7. Install NDT server on NETTEST5
  8. IEPM-BW Web Services - Yee
    1. Does our web services access work (need to contact Warren, await proposals, and stability of implementations) - Yee
  9. Set up Wiki
  10. Presentations/Talks/Visits/Papers/Documentation
  11.  IPv6
  • No labels