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

20030908: LDM - Linux meteop 2.4.18-3 #1 Thu Apr 18 07:37:53 EDT 2002 i686 unknown - LDM problem with the CLOCK ( ?) queue is empty .

>From: "Arline Bourdages" <address@hidden>
>Organization: Hydro Quebec
>Keywords: 200309081612.h88GCDna008002 LDM SIGALARM pqing

Hi Arline,

>Institution:  / Hydro quebec
>Package Version: ldm-5.2.2   on redhat  7.3
>Operating System: Linux meteop 2.4.18-3 #1 Thu Apr 18 07:37:53 EDT 2002 i686 u
> nknown
>Hardware Information: model name: Intel(R) Pentium(R) III CPU family 1400MHz
>cpu family      : 6
>model           : 11
>model name      : Intel(R) Pentium(R) III CPU family      1400MHz

>Inquiry: There is no message coming from the queue:
> I keek receiving the message: ( from lpqact 
>SIGALRM
>        End of Queue
>        SIGALRM
>        End of Queue
>        SIGALRM
>        End of Queue
>
>from the programe:  pqact -v -x -l /usr/local/ldm/pq.log 
>
>My time is : UCT  Is that ok?

Yes, this is normal.  You are seeing this particular message because of
the debug flag, '-x', being set.

What is going on is:

- pqact reads products from the product queue
- when pqact comes to the end of the queue (i.e., there are no more
  products to process), it sets calls pqsuspend
- pqsuspend sets a timer and pqact goes to sleep.  If a new product(s)
  is received before the timer goes off, pqact is sent a CONTinue signal
  (SIGCONT) telling it to wake up and process the new product(s).  If
  no new product is received by the time the timer goes off, an ALRM
  signal is sent to pqact.
- if pqact is running in debug mode, it will log the receipt of the
  ALRM signal and say why it got it

So, if you want to eliminate the debug messages from your log file,
you should edit ~ldm/etc/ldmd.conf and remove the debug flag from
the pqact invocation:

change:

exec    "pqact -v -x -l /usr/local/ldm/pqact.log -p .* "

to:
exec    "pqact -v -l /usr/local/ldm/pqact.log -p .* "

Also, the default pattern for pqact is '.*', and you can combine the
'-v' and '-l' flags so that your ldmd.conf entry would look like:

exec    "pqact -vl /usr/local/ldm/pqact.log"

This will assume that the LDM queue is in /usr/local/ldm/data/ldm.pq.

>meteop(arline)/proc: date
>Mon Sep  8 16:08:06 UTC 2003
>meteop(arline)/proc: 
>
>I do use ntp which 3 servers. 

OK, so your clock should be correct.

>The jobs are running:
>140 S ldm       1493     1  0  75   0    -   421 schedu 15:42 ?        00:00:0
> 0 rpc.ldmd -q /usr/local/ldm/data/ldm.pq /usr/local/ldm/etc/ldmd.conf
>100 S ldm       1494  1493  0  75   0    -   408 rt_sig 15:42 ?        00:00:0
> 0 pqbinstats
>100 S ldm       1495  1493  0  75   0    -   414 rt_sig 15:42 ?        00:00:0
> 0 pqact -v -x -l /usr/local/ldm/pqact.log -p .*
>100 S ldm       1496  1493  0  75   0    -   421 schedu 15:42 ?        00:00:0
> 0 exp -v -f EXP -b 4800 -p even -l /usr/local/ldm/logs/data.log /dev/ttyS0

Is 'exp' is a link to 'pqing'?  The standard LDM installation does not
have this link, however.  Here is a snippit from the pqing man page
that refers to this:

% man pqing

User Commands                                            PQING(1)

NAME
     pqing, dds, pps, ddplus, ids,  hds,  afos  -  product  queue
     ingester for Unidata LDM

SYNOPSIS
     pqing [-v] [-r rawfile] [-l logfile] [-f feedtype] [-b baud]
          [-q q_path] [-p parity] [-c] [-n] [-5] feedname

DESCRIPTION

 ...

     As a convenience, the program changes its default baud  rate
     based  on its name (argv[0]). For example, if the program is
     invoked as ids , the baud rate set to 9600.  For  afos,  the
     default  baud rate is 4800.  The program name also effects a
     ``feedtype'' field in  the  product  data  structure,  which
     helps  to  uniquely identify a data product. Conventionally,
     ddplus, ids, hds  and afos  are links to pqing .


>I did succes in writing in a file after using ntpdate <server>
>But I can not doing it again.

So, perhaps the problem is that your 'exp' process (presumably pqing)
is not reading anything from /dev/ttyS0, not that pqact is not processing
data that is being read in.   Verify that this is true by using 'notifyme':

notifyme -vxl- -f EXP -o 3600

If 'exp' (pqing) is reading data from /dev/ttyS0, then 'notifyme' will
list back the products that were received.  Here is an example
from a machine that is getting data from an upstream LDM:

% notifyme -vxl- -f EXP -o 3600

 ...
Sep 10 21:19:18 notifyme[71300604]: 86f380fe6343320783fbdd5be05ed21b      175 
20030910211918.067     EXP 124  
rtstats-6.0.10/chisel.rap.ucar.edu/CRAFT/kgldrrv_v_thelma.ucar.edu
Sep 10 21:19:18 notifyme[71300604]: 034b3597071693b59e4d73c9a9dd40c5      175 
20030910211918.067     EXP 125  
rtstats-6.0.10/chisel.rap.ucar.edu/CRAFT/ldmlsx_v_thelma.ucar.edu
Sep 10 21:19:18 notifyme[71300604]: eb636e2668fda95fab2cf54b4f7f3b43      190 
20030910211918.361     EXP 639  
rtstats-6.0.11/kepler.rwic.und.edu/NLDN/striker.atmos.albany.edu_v_adiabat.rwic.und.edu
 ...

If 'notifyme' shows that products are getting inserted into the LDM queue,
then we need to compare the headers for the products against the patterns
you have in your pqact.conf file.  If 'notifyme' shows no products being
inserted into your queue, the the problem is either in the feed or
in the 'exp' invocation in ldmd.conf.

>Help please!!

Please run 'notifyme' and let us know the results.

Tom Yoksas