<<< Chronological Index >>>    <<< Thread Index >>>

Re: TTM to-do list




> 3.5.  Upgrade SSH                                             ??     
>       This is waiting for ops to upgrade ssh @ ncc.

I think we closed this after Olaf's conclusion we were not affected?

> 4.2.  Implementation of RVEC and VBYE storage (TracerouteDB)  RW+S/W 8/00 

Done.

> 4.3.  CGI-interface to RVEC and VBYE storage  (TracerouteDB)  RW+S/W 8/00 

almost done; we have a pending request on taking configuration from
LIST_OF_TESTBOXES instead of hard-coding; will check with Manuel
(he was ill when Fotis send the last mails on the subject).

> 4.8.  Move daily plots to ROOT                                RW     8/00 

Done.

> 4.9.  Investigate what NTP error estimates can do for DQM     RW     ?? 

Still open. Preliminary conclusion is that the errors are not of much 
use when xntp has distant external peers such as osdorp configured; otherwise
might be able to catch the occasional "reboot/reset" effect, where NTP
slowly converges to the exact time.

> 4.10. Access to configuration and experimental parameters 
      from within ROOT (Related to (a.o.) 3.7, 3.10, 4.9))    ??

Open. Have been thinking about this, though. The new ROOT should
be able to talk to MySQL DB; we could upgrade the daily 'sattelites'
job, to a program that stores percentiles of satellite distribution
in a DB (either per day, or per 6hour interval); after defining
what "good" sat conditions are (2.5 percentile >= 1?) we could add it
to analysis code.

As for access to the configuration of the TTM network, we have an (at least
interim) solution in the form of the 'ttconfig' command; a perliminary C++
TestBoxConfig class provides abstract interface to the analysis software.

> 4.11. Plots on demand/CGI interface to daily plots            RW     

Alpha version ready at iii/cgi-bin/pod Improvements needed in cache
maintenance and load management (don't want too many simultaneous 
requests running). Probably want to turn prove-of-concept backend
shell script on ginkgo into real program as well.

> 4.12  df checks on our disks                                  RW     11/00 

[specifically for the 'root data' filesystems]

Open. Current root_data fs is 60% full, will certainly take us
through november and perhaps even a little longer. Deadline
for automated checks: St.Nicolaas's birthday.

> 6.4.  Summarize daily plots                                   RW     8/00

Open. Summaries are now on the plots only. Delay and loss parameters
should be extracted and stored in either ROOT or MySQL database.
Relates to 6.6.

> 6.6.  Turn Johann's code into production code                 ?? 

Most likely a complete rewrite is needed (not discarding Johann's work,
but turning the algorithms into better code).  The first thing to do
is efficient storing of delay and loss parameters (Johann used formatted
ascii as it was easiest and allowed him to focus on the analysis part).

--- NEW (PROPOSED) ---

6.9 graphical summaries					RW  12/00

In 6.4 a lot of numbers will be created for each individual box:
2.5, median and 97.5 percentile for delay and loss of each target.
The "scores" of item 6.5 are expected to summarize these into
one single number.

When a box has a 'bad' score, users might be interested in learning
more about the reasons for that; they may want to go one level deeper
in the data. My idea is to visually summarize the per-box numbers,
such that at one glance you get a feeling of the performance of
a box in the TTM network.

Similarly, for those interested in the status of TTM network
as a whole, a graphical representation of all scores may be nice.
(animals count "1, 2, 3 ... many" [*]  and although human brain
is more capable, it is impossible to compare scores of >40 boxes
in a glance)

I have some ideas on this, and want to work on a prototype in the
next weeks.

> 7.3   Add information on analyzing with root for TB hosts to
>       the website.                                            RW/HU  In Prgs

I'm in the process of wrapping up the delayplots code such that
it can be put on the website as an example of TTM data analysis
with ROOT. Will update and complete what we have on the website 
regarding this topic.

> 9.2.  Upgrade to the latest ROOT version                      RW    9/00

almost Done: I expect to switch versions tomorrow.


-- Rene


[*] http://www.sigmaxi.org/amsci/articles/00articles/hauserone.html





<<< Chronological Index >>>    <<< Thread Index >>>