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

20020726: Help! LDM 5.2 keeps crashing on Linux RH 7.3! (cont.)



>From: Gilbert Sebenste <address@hidden>
>Organization: NIU
>Keywords: 200207261442.g6QEgK915724 LDM 5.2 RedHat 7.3 Linux

Gilbert,

>A friend of mine who is the sysadmin at COD alerted me to this. Could it 
>be that I need to feed from another LDM 5.2?

You shouldn't have to, but ...  Anne ran extensive tests of the new
distribution on her RedHat 7.1 Linux system before releasing the code.

>Maybe this explains why the 
>other two don't die. I wonder what would happen if I fed from an LDM 5.1.4 
>on all 3 machines?

It seems like you could test this concept.  Try moving weather2 back to
using LDM 5.2 and one of your other machines to 5.1.4.  Then have the
other machine be the primary feed at your site and relay to weather2.
If weather2 stops crashing and the other machine starts crashing, it
would lend credence to the notion that 5.2 has problems feeding from
5.1.4 under RedHat 7.3 Linux.

By the way, I would ordinarily let Anne field these inquiries, but she
is out of the office at the moment.  Plus, she will be holding an LDM
workshop on Thursday-Saturday next week.

>>From address@hidden Fri Jul 26 16:51:20 2002
>Date: Fri, 26 Jul 2002 16:48:33 -0500 (CDT)
>From: David B. Bukowski <address@hidden>
>To: Gilbert Sebenste <address@hidden>
>Subject: weather 07/26/02:11.02 system check (fwd)
>
>i see this nearly every hour of your ldm stopping here is that something u
>know about?
>-dave
>
>
>------------------------------------------------------------------------------
> -
>David B. Bukowski      |email (work):          address@hidden
>Network Analyst                |email (personal):      address@hidden
>College of Dupage      |webpage:       http://www.cshschess.org/davebb/
>       
>Glen Ellyn, Illinois   |pager:                 (708) 241-7655 
>http://www.cod.edu/    |work phone:            (630) 942-2591
>------------------------------------------------------------------------------
> -
>
>---------- Forwarded message ----------
>Date: Fri, 26 Jul 2002 06:02:01 -0500
>From: root <root>
>To: root
>Subject: weather 07/26/02:11.02 system check
>
>Security Violations
>=-=-=-=-=-=-=-=-=-=
>Jul 26 10:39:54 weather weather2(feed)[17251]: pq_sequence failed: Input/outpu
> t error (errno = 5) 
>Jul 26 10:39:55 weather weather2(feed)[22199]: pq_sequence failed: Input/outpu
> t error (errno = 5) 
>Jul 26 10:40:32 weather weather2(feed)[22200]: pq_sequence failed: Input/outpu
> t error (errno = 5) 

Tom

>From address@hidden Sat Jul 27 08:44:32 2002
>Subject: Re: 20020726: Help! LDM 5.2 keeps crashing on Linux RH 7.3! (cont.)

Tom,

> >A friend of mine who is the sysadmin at COD alerted me to this. Could it 
> >be that I need to feed from another LDM 5.2?
> 
> You shouldn't have to, but ...  Anne ran extensive tests of the new
> distribution on her RedHat 7.1 Linux system before releasing the code.

Well, guess what. When I went over to LDM 5.1.4 on weather2 yesterday 
afternoon, yes what happened to weather last night? :-(
 
> It seems like you could test this concept.  Try moving weather2 back to
> using LDM 5.2 and one of your other machines to 5.1.4.  Then have the
> other machine be the primary feed at your site and relay to weather2.
> If weather2 stops crashing and the other machine starts crashing, it
> would lend credence to the notion that 5.2 has problems feeding from
> 5.1.4 under RedHat 7.3 Linux.

See above.
 
> By the way, I would ordinarily let Anne field these inquiries, but she
> is out of the office at the moment.  Plus, she will be holding an LDM
> workshop on Thursday-Saturday next week.

OK. But I think I have some evidence that this is true: a segmentation 
violation occurs when feeding from a 5.1.4 machine to a 5.2 machine.
But, weather3 didn't die...however, it doesn't get the Meteorologix radar 
feed, and it doesn't save anything from McIDAS (though it does get the feed).

My next test is to stop using the Meteorlogix feed, use the WSI NOWRAD(tm) 
feed instead, and see if that still crashes the machine. If it stops it
from happening, then that's the problem.

>From address@hidden Sat Jul 27 09:37:42 2002
>Subject: Re: 20020726: Help! LDM 5.2 keeps crashing on Linux RH 7.3! (cont.)

On Fri, 26 Jul 2002, Unidata Support wrote:

> You shouldn't have to, but ...  Anne ran extensive tests of the new
> distribution on her RedHat 7.1 Linux system before releasing the code.

OK. I now have two plausible theories why my LDM 5.2 is crashing.

1. LDM 5.2, on Redhat Linux 7.3, has problems with feeding from an LDM of 
version 5.1.4 or earlier.

2. LDM 5.2 is having problems with my Meteorologix MNG radar files coming 
in under whe "WSI" label under RedHat 7.3.

Occasionally, I would get a corrupted MNG file due to one reason or 
another, and it would crash 5.1.4. I now wonder if something was 
changed with the WSI receive protocol under LDM 5.2 to cause the LDM to 
crash.

So, here's what I did today to test out both theories.

1. weather2.admin is now back on LDM 5.2.
2. Weather.admin is now back on LDM 5.1.4.
3. Weather2 is currently NOT getting the MNG feed through weather, which 
it normally does.

We know that weather2 can't stay up for more than 12 hours (usually a 
lot less) with the configuration we had before (all machines on LDM 5.2).
If weather2 stays up all weekend, then we know it is the MNG files from
Meteorlogix causing the crashes. If THAT is the case, I can live, 
currently, with one machine only getting and processing the MNGs. But I 
really need two machines, so I'll ask for help or ideas on how to solve 
this. Three people use this feed; Duke Power, COD and myself.

If weather2 still crashes in its current configuration, then we know
that I have a plausable case for LDM-5.2 having trouble feeding from a 
5.1.4 machine under RedHat 7.3...since the 5.2 machines didn't crash when 
feeding from a 5.2 machine. Then, I'll ask if there's any way you can fix 
that.

Either way, by the end of this weekend, we'll know! Thanks for your help!

*******************************************************************************
Gilbert Sebenste                                                     ********
Internet: address@hidden    (My opinions only!)                     ******
Staff Meteorologist, Northern Illinois University                      ****
E-mail: address@hidden                                 ***
web: http://weather.admin.niu.edu                                      **
Work phone: 815-753-5492                                                *
*******************************************************************************
 > >Maybe this explains why the 
> >other two don't die. I wonder what would happen if I fed from an LDM 5.1.4 
> >on all 3 machines?
> 
> It seems like you could test this concept.  Try moving weather2 back to
> using LDM 5.2 and one of your other machines to 5.1.4.  Then have the
> other machine be the primary feed at your site and relay to weather2.
> If weather2 stops crashing and the other machine starts crashing, it
> would lend credence to the notion that 5.2 has problems feeding from
> 5.1.4 under RedHat 7.3 Linux.
> 
> By the way, I would ordinarily let Anne field these inquiries, but she
> is out of the office at the moment.  Plus, she will be holding an LDM
> workshop on Thursday-Saturday next week.
> 
> >>From address@hidden Fri Jul 26 16:51:20 2002
> >Date: Fri, 26 Jul 2002 16:48:33 -0500 (CDT)
> >From: David B. Bukowski <address@hidden>
> >To: Gilbert Sebenste <address@hidden>
> >Subject: weather 07/26/02:11.02 system check (fwd)
> >
> >i see this nearly every hour of your ldm stopping here is that something u
> >know about?
> >-dave
> >
> >
> >------------------------------------------------------------------------------
> > -
> >David B. Bukowski    |email (work):          address@hidden
> >Network Analyst              |email (personal):      address@hidden
> >College of Dupage    |webpage:       http://www.cshschess.org/davebb/
> >     
> >Glen Ellyn, Illinois |pager:                 (708) 241-7655 
> >http://www.cod.edu/  |work phone:            (630) 942-2591
> >------------------------------------------------------------------------------
> > -
> >
> >---------- Forwarded message ----------
> >Date: Fri, 26 Jul 2002 06:02:01 -0500
> >From: root <root>
> >To: root
> >Subject: weather 07/26/02:11.02 system check
> >
> >Security Violations
> >=-=-=-=-=-=-=-=-=-=
> >Jul 26 10:39:54 weather weather2(feed)[17251]: pq_sequence failed: 
> >Input/outpu
> > t error (errno = 5) 
> >Jul 26 10:39:55 weather weather2(feed)[22199]: pq_sequence failed: 
> >Input/outpu
> > t error (errno = 5) 
> >Jul 26 10:40:32 weather weather2(feed)[22200]: pq_sequence failed: 
> >Input/outpu
> > t error (errno = 5) 
> 
> Tom

>From address@hidden Sat Jul 27 09:59:16 2002
>Subject: Re: 20020726: Help! LDM 5.2 keeps crashing on Linux RH 7.3! (cont.)

Tom,

One more thing. I am suspecting the Meteorologix feed through the 
WSI protocol at this point. Weather3 doesn't get the MNG feed, has been on 
LDM 5.2 the whole time, and has yet to crash.

*******************************************************************************
Gilbert Sebenste                                                     ********
Internet: address@hidden    (My opinions only!)                     ******
Staff Meteorologist, Northern Illinois University                      ****
E-mail: address@hidden                                 ***
web: http://weather.admin.niu.edu                                      **
Work phone: 815-753-5492                                                *
*******************************************************************************


>From address@hidden Sat Jul 27 19:44:18 2002
>Subject: Re: 20020726: Help! LDM 5.2 keeps crashing on Linux RH 7.3! (cont.)

Tom,

Weather2 (running LDM 5.2) crashed when feeding from a 5.1.4 LDM 
(weather.cod.edu), without any Meteorologix radar mosaics being ingested. 
So, that rules that out. After weather.admin crashed this morning (then 
running 5.2), after feeding from weather2 (then on 5.1.4), there's no 
doubt in my mind about the incompatibility on RH 7.3, at least.
I'm sending a warning to the ldm-users group now. Oh, and weather3
has been pretty much been feeding from an LDM 5.2 machine, and no 
problems. So 5.2 feeding 5.2 has not been an issue...

*******************************************************************************
Gilbert Sebenste                                                     ********
Internet: address@hidden    (My opinions only!)                     ******
Staff Meteorologist, Northern Illinois University                      ****
E-mail: address@hidden                                 ***
web: http://weather.admin.niu.edu                                      **
Work phone: 815-753-5492                                                *
*******************************************************************************

>From address@hidden Sun Jul 28 08:36:22 2002
>Subject: Re: 20020726: Help! LDM 5.2 keeps crashing on Linux RH 7.3! (cont.)

Tom,

Just got word that Chris Novy, running 5.2 on a Sparc station using Sun OS 
5.8 is also having his LDM crash with queue corrupts.

*******************************************************************************
Gilbert Sebenste                                                     ********
Internet: address@hidden    (My opinions only!)                     ******
Staff Meteorologist, Northern Illinois University                      ****
E-mail: address@hidden                                 ***
web: http://weather.admin.niu.edu                                      **
Work phone: 815-753-5492                                                *
*******************************************************************************