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

FYI: created symlinks tt01 -> geuzenveld for old ROOT data


Up untill the last days of 1999 the primary name (as configured with
"hostname") for tt01 was geuzenveld.ripe.net; consequently, all files
created by TTM software had the word 'geuzenveld' in it. Around the
end of 1999, we decided to make tt01 the one and only name for this
testbox to reduce the number of exeptions in the naming scheme
(which is: box with id=XY gets hostname ttXY.ripe.net).

However,  since the configuration "DB" doesn't know of this change and
returns tt01.ripe.net when asked for the boxes which were online in 1999,
plots-on-demand would not find any data for this box for the times it went
by the name of geuzenveld.ripe.net.

Fotis suggested renaming all existing *geuzenveld.ripe.net* data files
to *tt01.ripe.net*; however, as raw data and ROOT data files already
made it to tape with the name geuzenveld and are recorded like that
in the online tape contents archives, I don't think that's a good idea:
any reprocessing, refetching files from tape will bring back the
old name. Instead I've now created symlinks in the appropriate 1999
root_data directory trees, such that plots-on-demand will find all
data for tt01.ripe.net.

It would probably be nice if sometime in future we could avoid the
convolution of testbox ids and names (e.g. it's pretty confusing if
tt98.ripe.net plots appear under www/test-traffic/Plots/tt05),
but for now it seems best to follow the plain and simple rule
hostname=tt<BoxID>.ripe.net for all machines participating in 
the measurement chain. Also: once datataking started, a hostname
should ALWAYS be the same for the duration of a box' active TTM life.

-- Rene





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