Re: LDM 5.1.4 - stops transfering from 1 site, while talking to others?

Anne Wilson wrote:
> Joe Van Andel wrote:
>
> > With LDM 5.1.4 on Redhat 7.2, we've seen instances where data stops
> > flowing from 1 site, while still flowing from other sites.
> >
> > For example, yesterday, syrah.atd.ucar.edu stopped getting our NEXRD2
> > feed from iita.rap.ucar.edu, even though the network was up and
> > iita.rap.ucar.edu was still up.  At the same time, syrah.atd.ucar.edu
> > was still receiving S-Pol gridded radar data, via linus.atd.ucar.edu
> >
> > The data feed resumed once we restarted ldm on syrah.
> >
> > Clearly, we don't want to have to monitor whether our data is flowing
> > and manually restart ldm if the data isn't arriving.
> >
> > Has anyone else seen network error conditions that can only be resolved
> > by restarting ldm?
> >
> > --
> > Joe VanAndel
> > National Center for Atmospheric Research
> > http://www.atd.ucar.edu/~vanandel/
> > Internet: address@hidden
>
>
> Hi Joe,
>
> Was there anything in your log or iita's log when this occurred?  It
> would show if there was an actual disconnect.  I'm wondering if this
> might be related to the work the UCAR network people are doing these
> days in upgrading routers, although I don't know of anything that
> happened specifically yesterday...

I don't have access to iita's log - and didn't see anything in my log.

> Has this happened more than once?

Yes.

> Regarding RH 7.2 vs. 7.3, perhaps Gilbert is right.  We've not
> experienced this problem on our own 7.2 machine, but we haven't pushed
> it much either.  We did, however, have another site that had trouble
> with 7.2 that was fixed with going to 7.3.  Are you considering
> upgrading the OS?

Sure, but since we're in the middle of the project, and 
syrah.atd.ucar.edu is a key machine, nobody will be very enthusiastic 
about taking the machine down to upgrade it.

> Also, when this occurs you could run 'netstat | grep iita' on syrah to
> see what state syrah thinks the connection is in.

Good idea, we'll try this if it happens again.

> I'm sorry that I can't offer much help on this one.  A band-aid would be
> a script that tracked input files (if you're actually filing the data)
> that sends an email if there's a gap.  Let me know if you'd like
> something like that, although that merely shifts the problem from
> monitoring a data stream to monitoring email.
>
> If this happens again and I'm around I would be happy to log in look
> around if you would like me to.

Thanks, I appreciate the offer.

> Anne



--
Joe VanAndel    
National Center for Atmospheric Research
http://www.atd.ucar.edu/~vanandel/
Internet: address@hidden