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

20030513: traceroutes from UQAM to downstream sites (cont.)

Christian,

> To: Unidata Support <address@hidden>
> From: =?ISO-8859-1?Q?Christian_Pag=E9?= <address@hidden>
> Subject: Re: 20030513: LDM problem connecting to upstream hosts at UQAM 
> (cont.)
> Organization: UCAR/Unidata

The above message contained the following:

> Steve,
> 
> I did traceroute to my two other feeds and the connection is very good  
> too:
> 
> traceroute to flood.atmos.uiuc.edu (128.174.80.47), 30 hops max, 40  
> byte packets
>   1  132.208.133.240 (132.208.133.240)  4 ms  1 ms  1 ms
>   2  132.208.6.25 (132.208.6.25)  1 ms  1 ms  1 ms
>   3  132.208.254.241 (132.208.254.241)  7 ms  2 ms  3 ms
>   4  uqam-gw.uqam-net.risq.net (206.167.112.25)  2 ms  1 ms  1 ms
>   5  uqam-intrarisq.dmtrl-uq.risq.net (132.202.40.161)  1 ms  1 ms  1 ms
>   6  cmtrl-uq.risq.net (206.167.145.118)  14 ms (ttl=249!)  137 ms  
> (ttl=249!)  92 ms (ttl=249!)
>   7  c4-mon01.canet4.net (205.189.32.226)  2 ms (ttl=248!)  2 ms  
> (ttl=248!)  2 ms (ttl=248!)
>   8  c4-tor01.canet4.net (205.189.32.37)  15 ms (ttl=247!)  12 ms  
> (ttl=247!)  11 ms (ttl=247!)
>   9  c4-mren.canet4.net (205.189.32.73)  23 ms  21 ms  21 ms
> 10  mren-lsd6509-m10.startap.net (206.220.240.85)  32 ms (ttl=245!)  22  
> ms (ttl=245!)  24 ms (ttl=245!)
> 11  uiuc-vbns-net.gw.uiuc.edu (192.17.10.5)  43 ms (ttl=243!)  36 ms  
> (ttl=243!)  39 ms (ttl=243!)
> 12  uiuc-ibgp1-lnk.gw.uiuc.edu (128.174.0.185)  38 ms (ttl=53!)  35 ms  
> (ttl=53!)  32 ms (ttl=53!)
> 13  uiuc-fwexit-net.gw.uiuc.edu (128.174.0.196)  36 ms  41 ms  56 ms
> 14  t-core1o.gw.uiuc.edu (130.126.0.189)  49 ms (ttl=51!)  39 ms  
> (ttl=51!)  38 ms (ttl=51!)
> 15  flood.atmos.uiuc.edu (128.174.80.47)  51 ms (ttl=50!)  39 ms  
> (ttl=50!)  41 ms (ttl=50!)
> 
> traceroute to sunset.meteor.wisc.edu (144.92.130.6), 30 hops max, 40  
> byte packets
>   1  132.208.133.240 (132.208.133.240)  1 ms  1 ms  1 ms
>   2  132.208.6.25 (132.208.6.25)  1 ms  1 ms  1 ms
>   3  132.208.254.241 (132.208.254.241)  6 ms  13 ms  11 ms
>   4  uqam-gw.uqam-net.risq.net (206.167.112.25)  3 ms  4 ms  2 ms
>   5  uqam-intrarisq.dmtrl-uq.risq.net (132.202.40.161)  14 ms  3 ms  2 ms
>   6  cmtrl-uq.risq.net (206.167.145.118)  2 ms (ttl=249!)  2 ms  
> (ttl=249!)  2 ms (ttl=249!)
>   7  c4-mon01.canet4.net (205.189.32.226)  2 ms (ttl=248!)  2 ms  
> (ttl=248!)  2 ms (ttl=248!)
>   8  c4-tor01.canet4.net (205.189.32.37)  10 ms (ttl=247!)  11 ms  
> (ttl=247!)  11 ms (ttl=247!)
>   9  c4-mren.canet4.net (205.189.32.73)  21 ms  21 ms  21 ms
> 10  mren-lsd6509-m10.startap.net (206.220.240.85)  20 ms (ttl=245!)  20  
> ms (ttl=245!)  20 ms (ttl=245!)
> 11  144.92.201.61 (144.92.201.61)  35 ms (ttl=244!)  26 ms (ttl=244!)   
> 30 ms (ttl=244!)
> 12  144.92.128.206 (144.92.128.206)  35 ms  31 ms  30 ms
> 13  sunset.meteor.wisc.edu (144.92.130.6)  28 ms  74 ms  30 ms
> 
> So the question is maybe why the bad connection to vortex lead to the  
> exit(1) of one ldmd children, which caused my ldm to not be able to  
> recover and get back IDS|DDPLUS feed.

If a requesting LDM process terminates, then, typically, the data
that it was requesting will no longer be received from the associated
upstream host.

I've changed that logic so that the requesting LDM process won't
terminate.  This new logic is in the new LDM release 6.0.11, which you
should probably consider installing since some of your connections are
so bad.

> I redid a search in my LDM logs, and it seems that the connection to  
> flood was always ok, except once yesterday afternoon where it timed out.
> For sunset, I saw an entry with: NOTICE: requester6.c:198: Upstream LDM  
> died.
> This all happened at the sametime, at around 19 GMT May 12th.
> 
> So, in my opinion, the fact that one ldmd children exited with status 1  
> (because of bad connection problems with one of my alternate ldm sites  
> feed) is what caused all the problems. Why did one ldmd children died?

The requesting LDM processes terminated because it couldn't connect to
the upstream LDM in any fashion.

> Anyway, I will still try to get more info for my network people here to  
> see why connectivity to some US sites are not very good today and  
> yesterday.
> 
> Thanks again,
> 
> Christian

Regards,
Steve Emmerson