Migration to the SL6.3 System |
In preparation to the SL6.3 upgrade (to be done BEFORE the upgrade):
- make sure to have the most recent dfos tool versions installed;
IMPORTANT: Do not install the new verisons of pet.py and
utilPACK, as those tools contain new python interpreter line and will
only work after the upgrade;
- call the updated scoreHC for all HC reports, to fix a format bug in the the score html pages
(the new firefox is less tolerant)
- save the crontab via: crontab -l > myfile
- terminate firefox before the hand-over
Handover muc0X to OTS for the system upgrade.
After you get the machine back:
- before you initiate the crontab file again, please check for your $TMP_DIR to physically exist (by default it would disappear with a new OS!). If not, create it.
- run calChecker -F once, in order to be sure that the $TMP_DIR/CALCHECK/PREVIOUS folder is filled properly, otherwise the incremental calls would fail or deliver junk
- fgwrite.c: is used to create ancillary fits frames.
On other muc machines it is available at:
/scisoft/share/iraf/extern/fitsutil/bin.redhat/fgwrite.e
while on muc09 it is in
/scisoft/share/iraf/extern/fitsutil/bin.linux64/fgwrite.e
include
export FG_WRITE=/scisoft/share/iraf/extern/fitsutil/bin.linux64/fgwrite.e
in the .qcrc and source it.
- Python: in the new system python is delivered from a different place:
was - /scisoft/bin/python or /qcdp/bin/python
is - /opt/python/bin/python
- Extend the PATH to
export PATH=/opsw/util/python/bin:${PATH}:/opt/sybase/bin:/opt/msql2/bin/:/opt/python/bin/python
in
dfosCron,
autoDaily,
-
use
#! /opt/python/bin/python
as the first line in
dfosRLSmassage.py,
pet.py,
and in the python QC scripts; if you download now the new versions of
pet.py and utilPack, the python call lines are already sorted out;
NOTE, that dfosRLSmassage.py is used by calChecker, so it has to be modified before calChecker can be run.
- issue with the python has_key (related to the pyFITS version);
solution provided by Wolfgang;
sed one-liner to massage python scripts
for the has_key issue,
in case it occurs within an "if" or "if not" statement
cat qc_mstplt.py | grep has_key
if
proHDUs[catg_master][0].header.has_key(key):
if not
temp_HDU[0].header.has_key(key):
if
proHDUs[catg_master][0].header.has_key(param[1]):
cat qc_mstplt.py | grep has_key | sed "s,\(if
[not]*\)\(.*\)\(\.has_key(\)\([^)].*\)\([^)]*\)\().*\),\1 \4 in \2:,"
if key in proHDUs[catg_master][0].header:
if not key in temp_HDU[0].header:
if param[1] in
proHDUs[catg_master][0].header:
- modify .esorex/esorex.rc for the new pipeline version
NOTE, that there is only new version of the detmon recipe available -detmon-1.2.5.
- install the saved crontab
- modify the ganglia link to http://mucmp-new.hq.eso.org/ganglia/
in config.dfoMonitor
- in case you use the "display" application; it is available at /usr/bin/display
but for some reasons the command works only if you call it like '/usr/bin/display '.
To fix it set in e.g. ~/.pecs/misc-all.ali the line
alias display=/usr/bin/display
- in case your qc scripts use gnuplot; gnuplot is no longer provided under /scisoft/bin;
version 4.2 is under /user/bin and 4.6 is under /vlt/VLT2013/gnu/bin/
- in case you experience problems with scoreQC check
echo $LANG
if this is set to C fine; if not
export LANG=C
in e.g. ~/.qcrc
- from the experience of working with CRIRES and FORS2 on muc01
after the upgrade...
if you have scisoft in your path be very careful; Isabelle noticed some core
dumps while using /scisoft/bin/dfits;
the conclusion is we should use /opt/qfits/bin/dfits ; I can confirm that in case
of FORS2 it worked very well;