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

Re: 20020822: ldmd won't stay running

Hi John,

Thank you for the information.

John C Nordlie wrote:
> 
> >
> > Hi John,
> >
> > Yuck!  I hate it when stuff just stops working!
> >
> > On your FreeBSD system, please do 'kill -l'.  It should list all the
> > signals.  Then, counting starting from 1 (not 0), please let me know
> > what signal 6 is on your system.
> 
> Signal 6 is ABRT.
> 
> >
> > Have you looked at the system logs?  Anything interesting there?
> 
> Yep, nothing of interest.  FreeBSD doesn't seem to like logging
> stuff with LDM (the ldmd.log file is always empty.  The snippet
> I sent is from /var/log/messsages).  Other than the signal 6 lines,
> nothing else seems to be showing up.
> 

I have heard of this logging issue on FreeBSD.  I know of two other
sites that run FreeBSD.  Both have modified the 'exec' commands in
ldmd.conf and the call to rpc.ldmd in start_ldm in the ldmadmin script
to include the '-l /usr/local/ldm/ldmd.log' option to get ldmd, pqact,
and anything else to log to the proper log.  We don't yet support
FreeBSD as we do not have one (yet) so we've never solved the problem.

I strongly suggest making this logging change.  We just went through
this with one of these FreeBSD sites, and once the logging was working
we got several new pieces of useful information.

In the meantime, perhaps 'rpc.ldmd -x -v -l -' will give some more
information.  (Using both -x and -v can 
provide additional messages.)


> >
> > How many rpc.ldmds do you have running?  Is it always the same
> > connection that aborts?  Especially if it's only staying up for  5
> > minutes, try running in debug mode, e.g., from the command line, start
> > the ldm with 'rpc.ldmd -x' and please send me the results.
> 
> One snapstop of a 'top' output had 15 rpc.ldmd processes running,
> but the number fluctuates.  All of them die off, leaving only one
> nexradII decoder running.  There are 15 REQUEST lines in the
> ldmd.conf file.
> 
> 'rpc.ldmd -x' only returns a process number, and the log files
> remain unchanged.  Two rpc.ldmd processes show up in 'top', sucking
> up all CPU cycles, but I don't see anything in 'ldmadmin watch'
> when I run it that way.
> 
> >
> > Just for the record, my last two support questions where things
> > mysteriously broke were due to network engineers installing software to
> > limit file sharing.  (In these cases it was Packetshaper - you probably
> > saw my email about that).  It is the time of year for them to do such
> > things.  However, your symptoms sound different.  Still, it might be
> > useful to ensure no changes were made at that end.  Can you ask someone
> > on your campus if anything has changed in networking?  Does ldmping work
> > for you for longer than 5 minutes?
> 
> I'll check on these.
> 

There's really not much here to go on.  Would you allow me to log in to
your machine?  If that's ok, please either call me or otherwise encode
the necessary information.  My phone number is (303) 497-8677.

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