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

Re: motherlode slowness

Daryl Herzmann wrote:
> 
> Hi!
> 
> On Tue, 25 Sep 2001, Anne Wilson wrote:
> >
> >However, when I was sampling motherlode yesterday afternoon, about 19Z -
> >20Z, everything was on time.  And, I'm not seeing any problems in the
> >stats files.  At what times did you take your samples?
> 
>         Shady Memory + Log Files
> 
> 20Z     I get an automated email with concerns about high latencies
> 20:13Z  Log onto ldmarchive and with a ldmadmin watch notice 40 min
>         latencies.  I then run a notifyme command to motherlode and
>         notice the same latencies.
> 20:30Z  I restart LDM on ldmadmin just as a test.  Same problems.
> 20:46Z  I failover to sunshine.ssec.wisc.edu to see if it is faster,  same
>         latencies.
> 20:50Z  Fail back over to motherlode and look at high latencies for all
>         sites feeding from motherlode on the internet FOS topo status
> 21:10Z  Desperate for data, I failed over to zelgadis.geol.iastate.edu
>         which was getting timely data.  It feeds from UIUC, which I
>         thought was wierd since it feeds from motherlode, maybe it failed
>         over?
> 21:25Z  Check motherloade with a notifyme command and it was current!  So
>         I failed back over to motherlode
> 22Z     I get another email about high latencies and notice that the
>         notifyme to motherlode is showing high numbers again
> 22:21Z  Failed over to zelgadis and left it like that till the next
>         morning.
> 
>         Further note on what zelgadis was getting yesterday afternoon.
> The model data was very latent 40 min, but the IDS|DDPLUs was timely.  It
> was a mixed bag.
> 
>         I was probably doing something stupid and it was a Monday!  I will
> send support emails instantly when it happens next.  I figured that you
> were probably getting enough emails about the latency problems judging
> from the IDD status pages yesterday afternoon.
> 
> Thanks!
>         Daryl

Thanks for log, Daryl.  This is interesting and mysterious.  I'll add
this to my set of data points.

We were seeing some high latencies from our other top tier nodes as
well.  I'm thinking it's the network, but I'm having trouble confirming
that.

Anne


-- 
***************************************************
Anne Wilson                     UCAR Unidata Program            
address@hidden                 P.O. Box 3000
                                  Boulder, CO  80307
----------------------------------------------------
Unidata WWW server       http://www.unidata.ucar.edu/
****************************************************