My system is on 24/7. The other day I went to check my email and found that my
box was off. There was a brown out, but I have a UPS. I thought maybe my UPS
was going as it's been on 24/7 as well for about 4 years. Today I came to check
on a download and found my box off again. I had just left it about 2 hours
before after working for about 3 hours. There was no brownout today. Besides
the mysterious shutoff my system seems to be in perfect shape. I checked dmesg
after reboot which didn't help as dmesg is cleared and repopulated after a
reboot. I checked syslog, and messages for the 23rd and 24th and didn't find
anything usefull. (No errors or any other obvious failures for the reboot) I
also checked faillog, but it seems to be a binary file. Not quite sure why.

So, any ideas/suggestions for where to start trouble shooting.

Why is faillog a binary file? What can I do to "fix" it.

System specs;
PSU PC Power and Cooling Silencer 400 (Great PSU btw)
Motherboard MSI KT3 ARU Ultra (getting on in years, and besides these 2
incidents not a single problem)
AMD AthlonXP Barton 2800
2x80GiB WesternDigital
1x250GiB WesterDigital
HP dvd640 DVD Burner
Liteon 48x CD Burner

I was considering the fact it may be my BIOS battery, though how that would
effect a running system I don't know. My BIOS time was about 10 minutes fast.
In the research I've done on the web, fast system time can mean a dying BIOS
battery.

--
Best Regards,

Mike Reynolds

MikeReynolds wrote:
> So, any ideas/suggestions for where to start trouble shooting.

did your machine shutdown properly, or was the power just cut off?

> Why is faillog a binary file? What can I do to "fix" it.

man 5 faillog
man 8 faillog

not that it'l help you. man 5 faillog:

" faillog maintains a count of login failures and the limits for each
account."

> I was considering the fact it may be my BIOS battery, though how that would
> effect a running system I don't know. My BIOS time was about 10 minutes fast.
> In the research I've done on the web, fast system time can mean a dying BIOS
> battery.

OTOH a clock running 4 years without being reset could well have gained 10
minutes. BIOS clocks are hardly atomic. ;-)

--
Joost Kremers joostkremers@yahoo.com
Selbst in die Unterwelt dringt durch Spalten Licht
EN:SiS(9)

MikeReynolds <mreynolds@invalid.com> wrote:
> There was a brown out, but I have a UPS. I thought maybe my UPS was
> going as it's been on 24/7 as well for about 4 years. Today I came to
> check on a download and found my box off again. I had just left it about
> 2 hours before after working for about 3 hours. There was no brownout
> today.

Some UPSes switch to battery when they think that input power isn't good
enough. This doesn't necessary need to be a brownout, it could be only a
short spike or that input power doesn't look like a sine wave.

I have had UPSes which because of this in combination with a bad battery
have caused more downtime than if I wouldn't have had any UPS at all.
Today I would not buy an UPS unless it has some kind of built in test
which is able to warn me if the battery is getting bad, and I also want to
be able to replace the battery without having to shut down connected
servers.

Are you sure that the battery of the UPS is able to keep your systems up
if your UPS decides to run on battery?

> I also checked faillog, but it seems to be a binary file. Not quite sure
> why.

The faillog file is for keeping track of login failures. For more
information about how to read that file, see "man faillog".

regards Henrik
--
The address in the header is only to prevent spam. My real address is:
hc7(at)uthyres.com Examples of addresses which go to spammers:
root@variousus.net info@k-soft.se info@k-software.biz root@localhost

On 2005-11-24, Joost Kremers <joostkremers@yahoo.com> wrote:
> MikeReynolds wrote:
>> So, any ideas/suggestions for where to start trouble shooting.
>
> did your machine shutdown properly, or was the power just cut off?
>

No it wasn't shutdown properly. Upon reboot, reiser said not cleanly unmounted.
(Reiserfs for /home) Unfortuneatly I haven't been at the box for wither of the
mysterious shutdowns.

>> Why is faillog a binary file? What can I do to "fix" it.
>
> man 5 faillog
> man 8 faillog
>
> not that it'l help you. man 5 faillog:
>
> " faillog maintains a count of login failures and the limits for each
> account."
>

Well don't I feel like an ass lol I'll read more into it.

>> I was considering the fact it may be my BIOS battery, though how that would
>> effect a running system I don't know. My BIOS time was about 10 minutes fast.
>> In the research I've done on the web, fast system time can mean a dying BIOS
>> battery.
>
> OTOH a clock running 4 years without being reset could well have gained 10
> minutes. BIOS clocks are hardly atomic. ;-)
>

The system hasn't been up for 4 straight years The only time it gets shutdown
is for maintenance, new kernels. Also before I was 100% Slack for reboots into
Windows. That was probably damn near 2 years ago.

--
Best Regards,

Mike Reynolds

On 2005-11-24, Henrik Carlqvist <Henrik.Carlqvist@deadspam.com> wrote:
> MikeReynolds <mreynolds@invalid.com> wrote:
>> There was a brown out, but I have a UPS. I thought maybe my UPS was
>> going as it's been on 24/7 as well for about 4 years. Today I came to
>> check on a download and found my box off again. I had just left it about
>> 2 hours before after working for about 3 hours. There was no brownout
>> today.
>
> Some UPSes switch to battery when they think that input power isn't good
> enough. This doesn't necessary need to be a brownout, it could be only a
> short spike or that input power doesn't look like a sine wave.
>
> I have had UPSes which because of this in combination with a bad battery
> have caused more downtime than if I wouldn't have had any UPS at all.
> Today I would not buy an UPS unless it has some kind of built in test
> which is able to warn me if the battery is getting bad, and I also want to
> be able to replace the battery without having to shut down connected
> servers.
>

This UPS does have a battery test, and a "Bad Battery" indicator light, which
is _not_ on. It also has a self test, which I belive needs the UPS not in use
to use. I should have ran the test before rebooting this last time.

> Are you sure that the battery of the UPS is able to keep your systems up
> if your UPS decides to run on battery?
>

Yes this UPS has enough. In the place I lived before this, we had brownouts
quite frequently. Some times 3 to 4 times a day, it was that bad. Old apartment
shitty landlord, way to many electronics. This was the reason for initially
getting the UPS. I have also ran the system off the battery for during a
blackout for about 10 minutes iirc. It's a Belkin, but the exact model escapes
me at the moment. Unfortuneatly the Belkin website doesn't go back 4 years for
model information.

>> I also checked faillog, but it seems to be a binary file. Not quite sure
>> why.
>
> The faillog file is for keeping track of login failures. For more
> information about how to read that file, see "man faillog".
>
> regards Henrik

Thanks for that. Previous poster pointed that out as well. I feel like an ass
for even bringing it up lol.

Is there any other place to look for relevant information?

Thank you

--
Best Regards,

Mike Reynolds

MikeReynolds wrote:
> On 2005-11-24, Joost Kremers <joostkremers@yahoo.com> wrote:
>> MikeReynolds wrote:
>>> So, any ideas/suggestions for where to start trouble shooting.
>>
>> did your machine shutdown properly, or was the power just cut off?
>
> No it wasn't shutdown properly.

that indicates that it wasn't the ups that told it to shutdown, i guess. so
the power just gets cut off. the things that come to mind are a faulty
power supply, loose wires or contacts, at any rate a hardware problem.

> Upon reboot, reiser said not cleanly unmounted.

/var/log/messages should also contain a message from init that it is
shutting down (switching to runlevel 6, to be precise).

--
Joost Kremers joostkremers@yahoo.com
Selbst in die Unterwelt dringt durch Spalten Licht
EN:SiS(9)

On Thu, 24 Nov 2005 21:19:21 GMT, MikeReynolds <mreynolds@invalid.com> wrote:

>
>Is there any other place to look for relevant information?

Do the Admiral's test, cut mains power and observe? The batteries
may be weak at four years old, increased internal resistance and
reduced capacity.

Perhaps you can beg/borrow/hire a variac from somewhere to create
brownouts and observe system response.

Borderline crazy: do you leave a CRT monitor running on output of
UPS? If so, have you disabled auto degauss? I have an old UPS
here that dies when asked to supply the degauss current surge...


Another observation: is the server not on Internet, can it run ntpd
to keep the time right?

Grant.

On 2005-11-24, Grant <g_r_a_n_t_@dodo.com.au> wrote:
> On Thu, 24 Nov 2005 21:19:21 GMT, MikeReynolds <mreynolds@invalid.com> wrote:
>
>>
>>Is there any other place to look for relevant information?
>
> Do the Admiral's test, cut mains power and observe? The batteries
> may be weak at four years old, increased internal resistance and
> reduced capacity.
>

Well I did just that (well yanked the plug anyway) and it is my UPS battery.
I'm going to see about replacing the battery, or maybe see if there is a way
to revive this one.

> Perhaps you can beg/borrow/hire a variac from somewhere to create
> brownouts and observe system response.
>

I have no idea what a variac is, and imagine this is some cheeky humour. I'll
google variac when I get a chance.

> Borderline crazy: do you leave a CRT monitor running on output of
> UPS? If so, have you disabled auto degauss? I have an old UPS
> here that dies when asked to supply the degauss current surge...
>
>

No CRT anymore, I got a nice 21.5" LCD about 6 months ago. I love it.

> Another observation: is the server not on Internet, can it run ntpd
> to keep the time right?
>

This is a workstation, and it is on the internet. I did look into ntp, but
after I found out my BIOS clock was 10 minutes fast as well, I figured setting
up NTP would be like putting a bandaid on an infected wound.

I got the model number of this UPS,
Belkin model F6C650-usb
650VA 390W

Great UPS besides this battery dieing on me. It has served me well for 4+ years
of 24/7 use. I got it on sale for cheap. Belkin has a "blowout" sale about once
a year. Some stuff is upto 80% off. IIRC this UPS was around $600 and I got it
for $200. Only thing that "got my goat" was that they claimed *nix compatibilty
It even came with the software to monitor the UPS. Turns out my model is USB
and the *nix compatibilty is serial only. Was a bit pissed to say the least.

Thanks for the help

--
Best Regards,

Mike Reynolds

On 2005-11-24, Joost Kremers <joostkremers@yahoo.com> wrote:
> MikeReynolds wrote:
>> On 2005-11-24, Joost Kremers <joostkremers@yahoo.com> wrote:
>>> MikeReynolds wrote:
>>>> So, any ideas/suggestions for where to start trouble shooting.
>>>
>>> did your machine shutdown properly, or was the power just cut off?
>>
>> No it wasn't shutdown properly.
>
> that indicates that it wasn't the ups that told it to shutdown, i guess. so
> the power just gets cut off. the things that come to mind are a faulty
> power supply, loose wires or contacts, at any rate a hardware problem.
>

No the UPS didn't tell the box to shutdown. See my post prior to this one.
As for the PSU, it's close to brand new. During last years hurricanes, the PSU
let out the magic blue smoke. I sent it in to PCP&C and they refurbished it.
(Was still under warranty) When I flipped the UPS over to get the model info
I did notice that the plug wasn't fully into the back of it, so it may have
lost it's line to the plug. And as the UPS' battery is dead, that would do it.

>> Upon reboot, reiser said not cleanly unmounted.
>
> /var/log/messages should also contain a message from init that it is
> shutting down (switching to runlevel 6, to be precise).
>

That's the first thing I looked for, and there was nothing.
I'm off to see about replacing the battery.

Thanks for the help.

--
Best Regards,

Mike Reynolds

On Thu, 24 Nov 2005 22:52:47 GMT, MikeReynolds <mreynolds@invalid.com> wrote:

>On 2005-11-24, Grant <g_r_a_n_t_@dodo.com.au> wrote:
....
>I have no idea what a variac is, and imagine this is some cheeky humour. I'll
>google variac when I get a chance.

Variable AC transformer, toroidial transformer with a tapped output,
usually 0 to input mains voltage + 10%, useful for power supplies and
building power amplifiers -- I still have one over thirty years old

Mostly useless, occasionally invaluable for electronics / electrical,
way OT for aols )

>This is a workstation, and it is on the internet. I did look into ntp, but
>after I found out my BIOS clock was 10 minutes fast as well, I figured setting
>up NTP would be like putting a bandaid on an infected wound.

BIOS clocks are uncompensated, unadjusted drifty cheap thingies need
ntpd to tame them.
....
>Great UPS besides this battery dieing on me. It has served me well for 4+ years
>of 24/7 use. I got it on sale for cheap. Belkin has a "blowout" sale about once

UPS usually discharge battery at way too high a rate, shortens battery
life, you may be better off with larger battery outside the UPS case?

Grant.