Re: 20020204: gempak, ldm

Sitler Jeffrey L Civ AFIT/ENP wrote:
> 
> Hi Anne
> No that wasn't my intent.
> I have stopped ldm and got the message
> 
> stopping the LDM server
> kill: 7585: no such process
> LDM server stopped
> 
> then when I do ldmadmin start
> I get the message
> starting the LDM server...
> sh: /home/kramer3/users/ldm/ldmd.pid: cannot create
> 
> I have no idea what that means,
> sorry about this, I thought I was cruising right along with this stuff.
> 
> So my intent was not to stop ldm and not have it running, and yes I could
> use some help.
> Thank you
> Jeff
> 

Hi Jeff,

Here's some more info for you.  The ldm uses two files to flag its
state.   

The first is ~ldm/ldmd.pid, the file that 'ldmadmin start' was
complaining about above.  This file serves to indicate that the ldm is
running, but if the ldm died abruptly it easily could have been left in
a bad state.  That must have happened, because ldmd.pid exists even when
your ldm is not running:  

fujita{ldm}51: ls -lt ldmd.pid
-rw-rw-r--   1 ldm      112            5 Dec 13 07:19 ldmd.pid

If this is the case, you can simply remove the file.  Or, do 'ldmadmin
stop' twice, or do 'ldmadmin clean'.  I did this on your machine and got
the following:

fujita{ldm}53: ldmadmin stop
stopping the LDM server...
Feb 5 16:29:46 UTC fujita.afit.af.mil : make_lockfile: another ldmadmin
process exists
fujita{ldm}54: ldmadmin clean
fujita{ldm}55: ldmadmin stop
stopping the LDM server...
Feb 5 16:29:58 UTC fujita.afit.af.mil : make_lockfile: another ldmadmin
process exists

These commands did remove ldmd.pid.  In addition, they also revealed a
second problem - they are refering to the second file I want to tell you
about.  The file /tmp/.ldmadmin.lck is used by ldmadmin so that only one
ldmadmin runs at a time.  I see that this file exists, even though
ldmadmin is not running:

fujita{ldm}57: ls -l /tmp/.ldmadmin.lck
-rw-rw-r--   1 jsitler  staff          0 Feb  5 07:40 /tmp/.ldmadmin.lck

This again indicates that the LDM did not shut down gracefully.  The
problem here is that the file is owned by 'jstitler', not by 'ldm'.  It
looks like you must have run ldmadmin as yourself.  This will cause
problems.  

So, as 'jstitler' you must remove this file.  After you do that I
believe you'll be able to start the ldm as usual.  Be sure to do all ldm
related commands, like 'ldmadmin stop' and 'ldmadmin start' as user
'ldm' and not as 'jstitler'.

Please try this and let me know how it turns out.

Anne 

-- 
***************************************************
Anne Wilson                     UCAR Unidata Program            
address@hidden                 P.O. Box 3000
                                  Boulder, CO  80307
----------------------------------------------------
Unidata WWW server       http://www.unidata.ucar.edu/
****************************************************