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

Re: 20020305: moving the ldm.pq file?

Christopher Herbster wrote:
> 
> Hi Anne,
> 
> I guess I didn't want to bother you with my reasons ... for fear you'd think 
> less of me.  (-:
> 

Oh, please grant me greater benevolence than that!

> I was going to move the queue off of the data disk that is NFS exported.  I 
> had some kind of disk "error thingy" )-: where the ldm died, and another 
> system locked up that was busy animating sat images in garp (with auto-update 
> enabled).  This happened over a weekend (of course) and I was sick the Monday 
> that we "fixed it" - forced reboot of the locked-up system.  A day or two 
> later when I returned to life I tried to do a "ldmadmin delqueue" as I 
> wondered about its integrity and the ldm had been dead for some time.  The 
> command never finished and would not allow a kill.  Even shutdown failed 
> (wow).  The power button won the war.  (I hate doing that.)
> 

Yuck!

> I still have no idea "who" (NFS, ldm, garp, X, ???, etc) did what that 
> everyone didn't like, but it seems logical that I should move the queue to 
> another drive that stays local to supercell.  How's this sound?
> 
> Cheers,
> 
> CH
> 

It's not clear to me if your data disk was exported *to* or *from*
supercell.  If the disk is local, making that disk exportable to other
machines should not be a problem for the queue.  But if the original
disk was nonlocal and exported to supercell, then I don't see how that
worked! Being a memory mapped file, we've always said that the queue
should be local.  So, if that's not the case then yes, you should move
it.

If I was you, I would just delete the queue and start over to ensure
that the queue is in a clean state after the crash, rather than copying
the queue.  This will mean requesting products an hour old, but if
you've been down for over an hour it would do that anyway.  You can
always change that default if you want only current data.

If you're ever stuck like that, you can always just 'rm' the file
'ldm.pq' to delete the queue.  Sometimes after a crash there's other
'trash' lying around - try 'ldmadmin clean' to get rid of any state
files that might remain.

Let me know if you have any other questions.

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