[ydl-gen] Syslog question regarding mysterious shutdowns

[Derick Centeno]

On Thu, 27 May 2010 13:43:19 -0500
Nick Jones <nicholas-jones at uiowa.edu> wrote:

> Hi Derick,
> 
> First off thank you very much for the very thorough and prompt
> response.
> 
You're welcome.
> # uname -r
> Linux lcn 2.6.15-rc5.ydl.1custom #1 SMP Tue Aug 8 17:01:58 CDT 2006
> ppc64 ppc64 ppc64 GNU/Linux
> I compiled the kernel with special modules, namely an LSI fiber
> channel controller module.
Nice.
> # ps aux |grep syslogd
> root      2768  0.0  0.0  1824  640 ?        S    May24   0:02
> syslogd -m 0 root     22311  0.0  0.0  3964  712 pts/0    S
> 13:17   0:00 grep syslogd

I can understand the syslogd sleeping or needing to be restarted on my
system as it is a laptop; you however are running YDL as a server.  I
would think that you'd have lots of logs to examine from syslogd.  So
something is being missed.

As a former System Admin myself, I don't recall all the tricks of the
trade.  However, I did find some references which could be useful to
you.  Some of these references are useful for me also as I don't like
daemons sleeping when they should be telling me what is going on.

Here are some references which you can explore on your own.  I'll
assume you've already investigated man/info as these references build
upon them as well.

*http://publib.boulder.ibm.com/infocenter/wmbhelp/v6r0m0/index.jsp?topic=/com.ibm.etools.mft.doc/an04230_.htm

*http://www.linuxjournal.com/article/2926

*http://www.linuxjournal.com/article/5476

*http://www.linuxhomenetworking.com/wiki/index.php/Quick_HOWTO_:_Ch05_:_Troubleshooting_Linux_with_syslog

Of course, there are limits regarding what one can gather on the net
which is reliable.  It may be time to invest in old style System
Administration tomes addressing Linux System Administration in
sufficient detail to help you "drill down" to resolving the mystery
quickly.

> I tried running tail -f /var/log/messages >> log.txt &; so that all
> messages were hopefully logged, but got nothing helpful.  

The references I recommend above should be more useful.

> I appreciate the suggestions about the UPC, however I know for a fact
> that power has never been interrupted to the battery bacukp, the
> batteries are good, and that it will not shut down (it beeps loudly
> until the batteries die).  In the past I've observed the problem and
> the batteries were all full charge minutes after I went into the
> server room to check why the system went down. Recently a major power
> update was performed, and the problem has gone away (I suspect that
> there was "bad" power involved), so I'm waiting to see if it comes
> back.
> 
> I'll repost if I find it to be related to the version of YDL or the
> packages I'm using, but for now I'm going to assume it's not that.
> If you have any more suggestions regarding helpful system messages
> I'd be interested, but I'm hoping the problem is gone for good.
> 
> Thanks again
> 
> Nick

No problem Nick.  However, in my previous post I also wanted to share
with you the resources of two different vendors who have diverse
strengths in providing technical resources (for a price, of course)
which you may find useful.  I didn't mention them earlier as I was
pressed for time; I'll direct you to their respective webpages
discussing the UPS systems they offer:

*http://eshop.macsales.com/Search/Search.cfm?Ntt=apc&Ntk=Primary&N=0&Ns=P_Popularity|1

*http://www.blackbox.com/Store/Results.aspx/search-ups%60%60si_product

Blackbox is a no nonsense tech shop I worked with years ago; they
should be helpful in recommending more esoteric approaches should there
be need.