[daip] AIPS user data areas
computer support
help at astro.UCLA.EDU
Fri Jan 10 19:42:33 EST 2003
As we prepare to install the 31DEC02 version of AIPS, I was wondering if
you could advise me on how to configure this better. I am referring to:
>Date: Fri, 9 Aug 2002 16:17:26 -0600
>From: Eric Greisen <egreisen at cv3.cv.nrao.edu>
>To: Jean Turner <chicag at astro.ucla.edu>
>Subject: Re: help - AIPS
[...]
>The
>usual way one does things is to name data areas in some obvious way
>
>/DATA/PRIMATE_1
>/DATA/PRIMATE_2
>/DATA/BONITO_1
>
>and then set it up so that no disk area is required. If I remeber
>correctly, all disk areas on all machines are required in your
>environment and that will slow things down a lot.
>
>The file that describes all this is in $AIPS_ROOT/DA00/DADEVS.LIST or
>maybe DADEVS.LIST.site_name. This gives the names of the data areas
>and whether they are required or not. A matching list is in NETSP
>that controls the TIMDEST limit and user numbers allowed use of those
>disks. Note that that user # control is less needed when disks are
>named properly (via link files is fine) than in the sort of system I
>remeber at UCLA.
>
>But you should not lose permsision over a data area or lose access to
>a data set in this even if disk numbers are inserted - just change
>INDISK and there you are. Or set INDISK=0 and do an MCAT and UCAT.
>
>There is a stand alone program that will recover data files is a
>user's CA file is trashed but the rest of the data are there.
>To see if that is of use, actually cd to the area that should have
>your data and do an ls *.uuu* where uuu is your user number in
>extended HEX. It might help to cd to $AIPS_ROOT, then source
>LOGIN.CSH (or dot LOGIN.SH for bash) and say $CDTST. Then you can
>say
> EHEX UUUU
>where UUUU is your user number in decimal and it will tell you uuu in
>base 36. If your data are there then you can say (having done the
>LOGIN.CSH annd CDTST)
> RUN RECAT
>and answer the questions.
>
>I hope this helps. If not, send me the contents of your DADEVS.LIST
>and NETSP files and I the details of where you think the data ought to
>be incl the user number and I will make further suggestions.
>
>Eric
Here is our current NESTP file (minus the comments):
/usr/local/pkg/aips/DATA/CLOTHO_1 14. 240 5209
/usr/local/pkg/aips/DATA/PUBLIC_1 14. 0 0 0 0 0 0 0 0
/usr/local/pkg/aips/DATA/PUBLIC_2 14. 0 0 0 0 0 0 0 0
/usr/local/pkg/aips/DATA/PUBLIC_3 14. 0 0 0 0 0 0 0 0
/usr/local/pkg/aips/DATA/PUBLIC_4 14. 0 0 0 0 0 0 0 0
/usr/local/pkg/aips/DATA/MIMOSA_1 14.0 410 411 412 413 414 415 416 417
/usr/local/pkg/aips/DATA/MIMOSA_2 14.0 4201 4202 4203 4204 4205 4206 4207 4208
/usr/local/pkg/aips/DATA/MIMOSA_3 14.0 2001 2002 2003 2004 2005 2006 2007 2008
/usr/local/pkg/aips/DATA/MIMOSA_4 14.0 4211 4212 4213 4214 4215 4216 4217 4218
/usr/local/pkg/aips/DATA/MIMOSA_5 14.0 1000 2000 3000 4000 5000 6000 7000 8000
/usr/local/pkg/aips/DATA/MIMOSA_6 14.0 1999 2999 3999 4999 5999 6999 7999 8999
/usr/local/pkg/aips/DATA/OSPREY_1 14. 0 0 0 0 0 0 0 0
/usr/local/pkg/aips/DATA/OSPREY_2 14. 0 0 0 0 0 0 0 0
/usr/local/pkg/aips/DATA/OSPREY_3 14. 0 0 0 0 0 0 0 0
/usr/local/pkg/aips/DATA/OSPREY_4 14. 0 0 0 0 0 0 0 0
/usr/local/pkg/aips/DATA/OSPREY_5 14. 0 0 0 0 0 0 0 0
/usr/local/pkg/aips/DATA/OSPREY_6 14.0 641 1977 1978 1979 1818
and the corresponding DADEVS.LIST (again without comments):
+ /usr/local/pkg/aips/DATA/PUBLIC_1
+ /usr/local/pkg/aips/DATA/PUBLIC_2
+ /usr/local/pkg/aips/DATA/PUBLIC_3
+ /usr/local/pkg/aips/DATA/PUBLIC_4
+ /usr/local/pkg/aips/DATA/CLOTHO_1
+ /usr/local/pkg/aips/DATA/MIMOSA_1
+ /usr/local/pkg/aips/DATA/MIMOSA_2
+ /usr/local/pkg/aips/DATA/MIMOSA_3
+ /usr/local/pkg/aips/DATA/MIMOSA_4
+ /usr/local/pkg/aips/DATA/MIMOSA_5
+ /usr/local/pkg/aips/DATA/MIMOSA_6
+ /usr/local/pkg/aips/DATA/OSPREY_1
+ /usr/local/pkg/aips/DATA/OSPREY_2
+ /usr/local/pkg/aips/DATA/OSPREY_3
+ /usr/local/pkg/aips/DATA/OSPREY_4
+ /usr/local/pkg/aips/DATA/OSPREY_5
+ /usr/local/pkg/aips/DATA/OSPREY_6
Each of the "HOST_#" above is a symbolic link to a filesystem on the remote
HOST.
Can you tell me what we should change to make this more optimal?
Also, suppose that one of the remote systems (say the "PUBLIC" one) is down
for a few days due to hardware failure. How will that affect the catalog
numbering and such that was worrying Jean?
Thanks for your help.
Philippe P. Brieu http://www.astro.ucla.edu/computing/
UCLA, Physics & Astronomy e-mail: help at astro.ucla.edu
8370 Mathematical Sciences phone: +1--310-206-8596
Los Angeles, CA 90095-1562 fax: +1--310-206-2096
More information about the Daip
mailing list