Page 1 of 4

A mysterious shut-down...

Posted: Sat Jul 29, 2006 7:18 am
by Oskar45
I normally leave my Fuel run 24x7; however, the other day when I came home it was down and after reboot, I'd found in syslog:
unix: |$(0x15a)WARNING: 001a01 ATTN: 1.5V low warning limit reached @ 1.340V.
unix: |$(0x15a)WARNING: 001a01 ATTN: 1.5V low warning limit reached @ 1.269V.
unix: |$(0x160)WARNING: 001a01 ATTN: 1.5V level stabilized @ 1.452V.
unix: |$(0x15a)WARNING: 001a01 ATTN: 1.5V low warning limit reached @ 1.255V.
unix: |$(0x160)WARNING: 001a01 ATTN: 1.5V level stabilized @ 1.354V.
unix: |$(0x158)WARNING: 001a01 ATTN: 1.5V low fault limit reached @ 1.199V.
unix: WARNING: Auto power down will be delayed until shutdown is complete.
unix: |$(0x163)WARNING: 001a01 ATTN: power down aborted, environmental monitor reset
Xsession: mephisto: logout
INFO: The system is shutting down.
INFO: Please wait.
4D:IRIS /usr/etc/eventmond[794]: The child process was killed by the signal 9
0D:IRIS inetd[265]: inetd received SIGTERM; terminating.
3F:IRIS syslogd: going down on signal 15

What in your opinion is the exact reason for that? Can I do anything about it?

Re: A mysterious shut-down...

Posted: Sat Jul 29, 2006 7:39 am
by hamei
Oskar45 wrote:I normally leave my Fuel run 24x7; however, the other day when I came home it was down and after reboot, I'd found in syslog:

Code: Select all

unix: |$(0x15a)WARNING: 001a01 ATTN: 1.5V low warning limit reached @  1.340V.
unix: |$(0x15a)WARNING: 001a01 ATTN: 1.5V low warning limit reached @  1.269V.
unix: |$(0x160)WARNING: 001a01 ATTN: 1.5V level stabilized @  1.452V.
unix: |$(0x15a)WARNING: 001a01 ATTN: 1.5V low warning limit reached @  1.255V.
unix: |$(0x160)WARNING: 001a01 ATTN: 1.5V level stabilized @  1.354V.
unix: |$(0x158)WARNING: 001a01 ATTN: 1.5V low fault limit reached @  1.199V.

What in your opinion is the exact reason for that?

Mainboard monitoring failure ...
Can I do anything about it?

You have a service contract. Get a new one. In the meanwhile < l1cmd env off > will get you back running. That's what us normal mortals have to do :) The environment monitoring on Fuel sucks. SGI probably lost plenty of money by cheapying out on the components there. Save a nickel, lose a dollar MBA-think. Oops.

Re: A mysterious shut-down...

Posted: Sat Jul 29, 2006 10:57 am
by Oskar45
hamei wrote:You have a service contract. Get a new one.

Heh, hamei, you really recall that I've a service contract!!! Congrats to your brain - should be preserved in some nicely-labelled jar!!! Anyway, I'll be fine with a new board in a few days. Don't want to think about what it oost me without contract, though!

Re: A mysterious shut-down...

Posted: Sat Jul 29, 2006 11:27 am
by hamei
Oskar45 wrote:
hamei wrote:You have a service contract. Get a new one.

Heh, hamei, you really recall that I've a service contract!!! Congrats to your brain - should be preserved in some nicely-labelled jar!!!

It was, originally. Abby something ...

Posted: Mon Jul 31, 2006 6:51 am
by zafunk
I had the same problem recently and had to kill the environmental monitoring. I don't have a contract though :(

Re: A mysterious shut-down...

Posted: Mon Jul 31, 2006 7:03 am
by unixmuseum
hamei wrote:It was, originally. Abby something ...
:-D :-D :-D

Posted: Mon Jul 31, 2006 7:26 am
by Oskar45
zafunk wrote:I had the same problem recently and had to kill the environmental monitoring.

Hmm, I'm not sure that killing env monitoring will save you. As I've noted in my original post of this thread, the messages are written by the kernel so they will be in the syslog even if eventmond is not running. Also, I suspect my box would have been killed anyway regardless of whether eventmond was running or not ...

Posted: Mon Jul 31, 2006 7:36 am
by zafunk
Oskar45 wrote:Hmm, I'm not sure that killing env monitoring will save you. As I've noted in my original post of this thread, the messages are written by the kernel so they will be in the syslog even if eventmond is not running. Also, I suspect my box would have been killed anyway regardless of whether eventmond was running or not ...


Hmm.... well, only time will tell. So far, turning the monitoring off has saved me, but I may have to get a new mobo eventually :(

Posted: Mon Jul 31, 2006 8:40 am
by nekonoko
Oskar45 wrote:
zafunk wrote:I had the same problem recently and had to kill the environmental monitoring.

Hmm, I'm not sure that killing env monitoring will save you. As I've noted in my original post of this thread, the messages are written by the kernel so they will be in the syslog even if eventmond is not running. Also, I suspect my box would have been killed anyway regardless of whether eventmond was running or not ...


If you're shutting off env monitoring on the L1 (which is the case here) it won't report anything to the kernel. No-one said anything about turning off eventmond - this is more low level :)

Posted: Mon Jul 31, 2006 5:56 pm
by hamei
zafunk wrote: Hmm.... well, only time will tell. So far, turning the monitoring off has saved me, but I may have to get a new mobo eventually :(

Going on two years now here .... if it finally dies it'll go into the scrapper. The Fuel just isn't that great a computer to spend SGI's version of money on.

Posted: Mon Jul 31, 2006 10:18 pm
by yetanother**ixuser
agreed :twisted:

Posted: Tue Aug 01, 2006 11:06 am
by foetz
hamei wrote:
zafunk wrote: Hmm.... well, only time will tell. So far, turning the monitoring off has saved me, but I may have to get a new mobo eventually :(

Going on two years now here .... if it finally dies it'll go into the scrapper. The Fuel just isn't that great a computer to spend SGI's version of money on.


yes, sadly the fuel was more sort of a mips-pc like the alpha-pc those days or the 604e based ibms.

Posted: Tue Aug 01, 2006 5:15 pm
by hamei
foetz wrote: yes, sadly the fuel was more sort of a mips-pc like the alpha-pc those days or the 604e based ibms.

Didn't mean to come off as quite so negative about the Fuel - it's not a bad computer, actually. It's just that SGI still had their heads up their asses when they priced it. Okay, the mainboard and cpu are low-volume high-cost items. But $600 for an off-the-shelf peecee case ? The exact same one they used in the 230 and 330 machines ? And the rest of their Fuel prices are equally nonsensical. No wonder they are bankrupt. No one who isn't a fanboy (us) is gonna spend six times what something is worth just because it says SGI on the outside. And then they shit on their fanboy constituency. Great.

Posted: Tue Aug 01, 2006 5:30 pm
by Dr. Dave
You know, it sounds like these monitoring boards can probably be fixed... judging by what Oskar posted earlier, looks like it may be capacitors or something oscillating in the buffer amps. Anyone got a really good clear picture (IC part number readability is good) of what one looks like, working or not?

Posted: Wed Aug 02, 2006 6:15 am
by Oskar45
nekonoko wrote:
Oskar45 wrote:
zafunk wrote:I had the same problem recently and had to kill the environmental monitoring.

Hmm, I'm not sure that killing env monitoring will save you. As I've noted in my original post of this thread, the messages are written by the kernel so they will be in the syslog even if eventmond is not running. Also, I suspect my box would have been killed anyway regardless of whether eventmond was running or not ...


If you're shutting off env monitoring on the L1 (which is the case here) it won't report anything to the kernel. No-one said anything about turning off eventmond - this is more low level :)

Hmm - does that mean, if you shut off monitoring on the L1, the box would not die even if the voltage gets too low???