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

20000202: more on testing mcidas

>From: David Fitzgerald <address@hidden>
>Organization: Millersville University of Pennsylvania
>Keywords: 200001262144.OAA16099 McIDAS-X 7.6

Dave,

>FYI 
>I rebooted twister and tried to run the tests on mcidas.  This time I got
>these messages after starting mcidas:
>
>/software/mcidas% mcidas
>WARNING: only run as the user 'mcidas' for supervisory tasks
>Removed Graphics Memory Object
>/software/mcidas/.mctmp/201/*: No such file or directory
>Could not clean up /software/mcidas/.mctmp/201

This at least shows progress.

>The supervisory warning is normal of course,

Right.

>but I have not seen the 
>Removed Graphics Object message before.

Graphics memory is shared memory resident.

The could not clean up message is related to the cleanup process, 'mccheck'
not being able to remove an NFS resource file in ~mcidas/.mctmp/201.

>Needless to say mcidas still hangs.

Bummer.

>Should I recompile or maybe increase shared memory?

I just logged in and am in the process of poking around.  I removed the
~mcidas/.mctmp/201 directory and its contents.  This is something that
I was unable to do before you rebooted.

You have plenty of shared memory configured for this system (the
recommended 512 MB), so you don't need to add more.

As far as recompiling, I am now installing the distribution:

<login as 'mcidas'>
cd mcidas7.6/src
make install.all

and will go from there.  I will be looking to see if anything jumps out at
me after things in their install locations.

OK, here is the bad news:

o after I installed McIDAS, then ROUTE PP BATCH invocations from the ldm
  account started firing off.  This has caused your 'ldm' user to have
  a number of hanging batch.k and nortmapr routines:

/software/mcidas/.mctmp% ps -u ldm
   PID TTY      TIME CMD
   303 pts/0    0:00 csh
   429 pts/0    0:13 pqutil
  5921 ?        6:06 dchrly
   403 ?       18:17 pqexpire
   405 ?       20:43 pqbinsta
   402 ?        0:01 rpc.ldmd
   406 ?       50:04 pqact
   407 ?       24:08 rpc.ldmd
   408 ?        0:14 rpc.ldmd
   409 ?        3:19 rpc.ldmd
   410 ?        0:04 rpc.ldmd
 27963 ?        0:03 dcuair
   432 pts/1    0:00 csh
 20383 ?        0:00 goescomp
 20346 ?        0:00 batch.k
  1766 ?        0:01 dcgrib
  1915 ?        0:00 dcnldn
 20334 ?        0:00 batch.k
 20330 ?        0:00 batch.k
 20508 ?        0:00 batch.k
 14337 ?        0:45 rpc.ldmd
 20382 ?        0:00 batch.k
 20536 ?        0:00 batch.k
 23259 ?        0:24 dcsynop
 20351 ?        0:00 batch.k
 20514 ?        0:00 batch.k
 20369 ?        0:00 batch.k
 20372 ?        0:00 nortmapr
 20539 ?        0:00 nortmapr
  1261 ?        0:05 dcsynop

  Since I do not have a login as 'ldm', I could not kill them.  To prevent
  more from appearing, I uninstalled the McIDAS binaries.

o the McIDAS stuff is hanging when trying to set file control information
  on a file called mcclean.lock (gets created in the subdirectories of
  .mctmp).  Mike Schmidt and I were looking at your system until he
  had to leave; we believe that the problem is centered on how /software
  is being mounted from snowball.

We traced the sequence of events leading to the hanging condition by
running:

truss ~mcidas7.6/src/mcenv

What we found was:

 ...
open("/software/mcidas/.mctmp/42704/TERMCHAR.001", O_RDWR|O_CREAT, 0600) = 4
write(4, "80", 1)                               = 1
close(4)                                        = 0
open("/software/mcidas/.mctmp/42704/GRAPHICS.KEY", O_RDWR|O_CREAT, 0600) = 4
write(4, "80", 1)                               = 1
close(4)                                        = 0
open("/software/mcidas/.mctmp/42704/STRTABLE", O_RDWR|O_CREAT, 0600) = 4
write(4, "80", 1)                               = 1
close(4)                                        = 0
open("/software/mcidas/.mctmp/42704/mcclean.lock", O_RDWR|O_CREAT, 0600) = 4
write(4, "80", 1)                               = 1
fcntl(4, F_GETFD, 0x00000000)                   = 0
fcntl(4, F_SETFD, 0x00000001)                   = 0

The last line is where things are hanging.  the file
/software/mcidas/.mctmp/42704/mcclean.lock was created, but the system was
unable to change something with the file, so it was waiting.  This must
have something to do with how the file system is being mounted.  Mike
said that he would try and take a closer look at things tonight from
home, but I feel certain that he will need root privilege to do anything
substantial.

Tom