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

20030521: A SIMPLE LDM REQUEST



>From: Tom Grzelak <address@hidden>
>Organization: Rutgers
>Keywords: 200305211638.h4LGcfLd028191 LDM ldmadmin IDD

Hi Tom,

>You know in all the years and changes I have seen in LDM there has been one 
>issue IMHO that has never be fixed.  Let's say I edit the pqact.conf and 
>restart the process.  LDM will restart with no problem.  ldmadmin watch 
>will show products streamin' by.  But if there is a single glitch in the 
>pqact.conf NO products get processed or ingested.

Actually this is not entirely true.  No pqact.conf entries starting at
the one with the error down to the end of the file will get processed.
The ones above the line with the mistake will get processed.

Now, if you modify ldmd.conf and have a typo, then you are correct, the
only place that this gets noted is in the LDM log file.  If this is
your point, I agree with you: the message should be put in the log file
AND sent to the person trying to start the LDM.

If you really are talking about pqact.conf, then our solution was to
provide the tool 'ldmadmin pqactcheck' that will check the integrity
of pqact.conf entries and tell you if there is something wrong.

>What amazes me is that 
>the error message that there is a problem ONLY gets logged to the 
>ldmd.log.  I feel it would be very beneficial that it also gets sent back 
>to the command line.  Since that is a major problem, it should be readily 
>seen without having to resort to log checking.

This really does sound like the ldmd.conf scenario, and I totally agree
with you.

>For example, last Friday was a typical Friday.  I am rushing to get out of 
>here and I see that I will run out of space on the weather server over the 
>weekend.  I edit the pqact.conf and remove a ingest line for a particular 
>product that I won't need this weekend.  I restart LDM and all seems 
>well.  I log on on Sunday and find I have ingested zero products for the 
>weekend because of a pqact.conf typo.

This should have only been the case if the entry you modified was at or
very near the top of your pqact.conf file.

>Can we have when we restart LDM, a message that flags a bad pqact.conf file 
>will appear if the file is bad.

I am CCing Steve Emmerson on this in order to pass along your, IMHO,
reasonable request.

>For example:
>
>LDM:/users/ldm/etc[7]% ldmadmin stop
>Flushing the LDM product-queue to disk...
>Stopping the LDM server...
>Waiting for the LDM to terminate
>Waiting for the LDM to terminate
>LDM server stopped
>
>LDM:/users/ldm/etc[8]% ldmadmin start
>Starting the LDM server (this can take a while)...
>
>WARNING: Your pqact.conf file has an error in it on line 325.
>No products will be ingested until you fix the problem and restart LDM.
>
>the LDM server has been started
>LDM:/users/ldm/etc[9]%
>
>I have had this happen a few times to me and it is very frustrating.  It 
>seems like it should be easy enough to do, especially since you log it.

For now, please run 'ldmadmin pqactcheck' every time you modify
pqact.conf.  Also, there is no need to stop and restart your LDM after
modifying pqact.conf.  If 'ldmadmin pqactcheck' indicates that your
pqact.conf file entries are syntactically correct, then you can tell
pqact to reread the file with 'ldmadmin pqactHUP'.

>Thanks!

Your request has been passed along!

Now, here is a request from us:

I see that you are still feeding NNEXRAD data from
snow.nrcc.cornell.edu.  Cornell is facing a pay-per-byte pricing change
for their internet use, so we are asking all sites feeding from them to
change that/those feed requests to atm.geo.nsf.gov.  This machine,
housed in the ATM offices of NSF in Arlington, VA, is already setup to
allow feed requests from you.

Thanks for your help in offloading Cornell's IDD machine!

Tom

>From address@hidden Wed May 21 11:06:16 2003

Thanks for your help.  I will make the data request change.

Tom