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

20020417: 20020417LDM shutdown problem (cont.)



>From: "Kevin Polston" <address@hidden>
>Date: Wednesday, April 17, 2002 0:52 am
>Subject: 20020417: recurring LDM shutdown problem 

Kevin,

>Here are my log files (ldmd.log and ldmd.log.1).  I see some things that
>don't look good. 

Very quickly I see indications that your product queue is not large
enough:

Apr 17 02:58:19 localhost 129.93.52.150[20251]: pq_del_oldest: no unlocked 
products left to delete!
Apr 17 02:58:19 localhost 129.93.52.150[20251]: comings: pqe_new: Resource 
temporarily unavailable

>I don't understand why something like this would happen especially since
>I had no problems previously.  Hope the log files help.

If the problem really is that the queue is too small, why it your system
would start having problems is understandable (the volume of the datastream
continues to grow), and the fix is easy:

1) shutdown your LDM
2) delete the queue file:

ldmadmin delqueue

3) edit ldmadmin and change the size of the queue that it can make:

cd ~/bin
<edit ldmadmin>

change the line that sets $pq_size:

$pq_size = 100000000;          <- the default sent out was 100 MB, a size that
                                  is way too small given the current contents
                                  of NOAAPORT

   Given what you are trying to ingest, and IF you have the disk space
   (check the file system that contains ~ldm/data to see how much space
   is free), I recommend changing the queue size to be well over 500
   MB; 1 GB would not be excessive:

$pq_size = 1000000000;

   You will have to adjust this number to be as large (up to 2 GB) as
   possible given the amount of disk space that you have.  Note that
   the queue does not grow, so the space needed for the queue today
   will be the same as for tomorrow, etc.  What does change, however,
   is the contents of the datastream (especially model output), so you
   have to keep an eye on your disk to make sure you don't fill up.

4) make the new queue (this will take some time, so be patient):

ldmadmin mkqueue

5) restart your ldm:

ldmadmin start

After deleting a queue, your system will request a full hour's worth of
data from its upstream host(s), so it will take time for you to catch
back up to the point of receiving current data with little delay.

Tom