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

Re: 20020607: comment from Larry Riddle regarding LDM connections

Unidata Support wrote:
> 
> Anne,
> 
> I wrote to Larry asking for a clarification of a comment he made about
> having to stay up and manually get his LDM to feed from thelma.
> The relevant part of what I asked and Larry's response are below.
> 
> Larry's observation about having to bounce the LDM to get data flowing
> again is contrary to my "understanding" of how the LDM should work.
> Am I missing something?
> 
> Tom
> 
> ------- Forwarded Message
> 
> >To: Unidata Support <address@hidden>
> >cc: Larry Riddle <address@hidden>
> >From: Larry Riddle <address@hidden>
> >Subject: Re: 20020607: Thelma Sick?
> >Organization: UCAR/Unidata
> >Keywords: 200206071341.g57DfZJ25167
> 
> >I am a little confused about your comment about having to stay up
> >to try and connect to thelma time after time.  Your LDM should do that
> >for you every 30 seconds or so.  Am I missing something obvious here?
> >Perhaps your LDM is exiting since it is trying to get all of its feeds
> >from one upstream host, and that host is refusing to feed?
> 
> All I can report is what I think I observed.
> 
> Thelma disconnected twice.  On both occasions, after waiting about 30
> minutes, I still got no data.  On the other hand, if I bounced the ldm
> (stopped and restarted) the ldm, data would flow immediately.
> 
> Caveat: That's what I think I saw, not necessarily what really
> occurred.  It was late here and it was the end of a long day.
> 
> Larry

Hi Tom and Larry,

In looking at Larry's logs, at 3:11Z he disconnected from thelma and
immediately reconnected.  This correlates with the Juniper problem Mike
told us about. (Larry, we've been going through router "upgrades" :-/
that have caused some temporary connectivity losses.)  After the
reconnect he got a few duplicate products, then things were fine.

By coincedence, at 4:02Z, he got that wierd assertion failure that he
gets now and then.  That failure causes the ldm to exit.  I'm sure this
confounded the picture for him.

At 4:36 the ldm was restarted.  It ran normally and was getting products
until 4:40 when there was what appears to be a normal shutdown.  At 4:42
it was restarted again.  I assume this was one of Larry's "bounces",
Larry, is that right?  If so, what caused you to do the restart?

So the LDM did what it was supposed to do when it lost the connection to
thelma.  On the other hand, this assertion failure is a sporadic,
ongoing problem, unique to aeolus, that I haven't been able to solve.  
I can't duplicate it, but now I'll take a look at yet another sample...  

Anne