[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]

20000913: ldm-mcidas-7.6.4 install and McIDAS-X 7.613 update

>From: Michael Keables <address@hidden>
>Organization: Geography Dept, University of Denver
>Keywords: 200009121648.e8CGmYb24740 ldm-mcidas PNG compressed images pnga2area

Mike,

>Both telnet and ssh have been enabled on cyclone to machines with
>hostnames .unidata.edu, so hopefully you'll be able to login.

OK.  I got your second note and have been merrily making modifications
to cyclone.  I will try to list all of the mods I made here, so you will
know what went on.

>As part of the security script (YASSP), the cron.allow file did not
>include any user except root. Users ldm, mcidas, mcadde have since been
>added. I've noticed that the /data directory is full, and I suspect that
>is due to cron not running mcscour as needed.

OK.  I added cron entries for data scouring as 'ldm'.  This includes
scouring of the McIDAS-XCD generated files which are going to /data/ldm/mcidas.

Here is what I did:

<login as 'ldm'>

<added some aliases to .cshrc to make my life easier :-)>

cd ~/ldm-mcidas-7.6.4/bin
cp area2png lwtoa3 nids2area nldn2md pnga2area proftomd uwgrid.sh ~/decoders
cd ~/decoders
rm lwfile lwtmd2 gunrv2 cdftomd
<edit batch.k, xcd_run, and uwgrid.sh to make sure that McIDAS environment
 variable settings were correct>
cd ~/etc
<edit ldmd.conf (and all of your various copies of it) and setup request
 lines for DDPLUS|IDS|HRS|FSL2 == WMO|FLS2 and MCIDAS:

request WMO|FSL2 ".*" cirrus.al.noaa.gov
request MCIDAS "pnga2area Q[01]" cirrus.al.noaa.gov
request NLDN    ".*" striker.atmos.albany.edu

<edit pqact.conf and fix the pnga2area entry:

MCIDAS  ^pnga2area Q. (..) (.*) (.*) (.*) (.*) (........) (....)
        PIPE    -close
        pnga2area
        -a /export/home/mcidas/data/SATANNOT
        -b /export/home/mcidas/data/SATBAND
        -d /data/ldm/mcidas -r \1,\2

o added cron entries for rotating log files and scouring data files

<as 'root'>

o add needed entries to /etc/syslog.con, /etc/rpc, and /etc/services

o killed syslogd and restarted it (it was not logging to ldmd.log).

  NOTE: Solaris 5.[67] has problems with syslogd hanging.  It may
  be necessary for you to kill and restart syslogd as 'root' in order
  to keep logging working.  Also, you should attempt to keep up
  on OS patches especially related to syslogd and ALL security
  issues!

o changed the group of files and links in ~ldm to be 'ldmgroup'.  It
  looks like 'ldm' was in the group 'staff' at one point and is now
  in 'ldmgroup'.  I was just trying to make everything have the same
  group.

o changed the ownership of the ~ldm/data link to 'ldm'; it was 'root'

<as 'mcidas'>

o upgrade your distribution from 7.612 to 7.613.  This was a minor
  upgrade.

o fixed REDRIECTions in the ~mcidas/workdata/LWPATH.NAM file.  There
  were numerous references to /var/data/mcidas, and that directory
  doesn't exist.  I changed all of these to /data/ldm/mcidas.

  fixed REDIRECTions in ~mcidas/data/LOCAL.NAM.  Same comment as for
  LWPATH.NAM.

o deleted the GBs of data in /data/ldm/mcidas that was really old.
  I also deleted a number of zero length files in /data/ldm/mcidas.
  After you ingest/decode for a couple of days, all truncated files
  should have been written over by new data.

o added ROUTE.SYS entries for the CIMSS products that were added to
  the UW broadcast at the end of June:

  cd ~/workdata
  ftp ftp.unidata.ucar.edu
    <user> anonymous
    <pass> your email address
    cd pub/mcidas
    get CIMSS.BAT
    quit
  batch.k CIMSS.BAT

  I also created a local copy of DROUTE.BAT in ~mcidas/data as per the
  online McIDAS installation instructions.  I added the CIMSS.BAT
  entries to the end of this copy, ~mcidas/data/LROUTE.BAT.  You can
  now easily recreate your routing table (ROUTE.SYS) if you need to.

As I write this, I am watching data roll in to cyclone and get decoded.
The change made to the MCIDAS request in ldmd.conf cuts down on the
amount of image data that is coming to your machine.  This change
would not be necessary as of this Friday, but I figured "what the hey".

Please let me know if you find anything that doesn't make sense, or that
is not working correctly.

Tom