[comp.os.minix] Minix 1.5.10 uptime/reliability

ghelmer@dsuvax.uucp (Guy Helmer) (07/14/90)

I'm hoping to set up my old AT&T PC6300 to run as a 24-hr mail and
public-domain server.  I have tried running Minix continuously on it
without rebooting, but after two days it started acting wierd
and after three it completely died (with very little system load!).

How long has everyone else managed to keep Minix up, and on what
systems?

I suspect an 80[23]86 might keep Minix running longer,
or point out problems better, than an 8086 or 68000, but I haven't
verified this.  I wish I could get kernel core dumps to find out
what crashed the system when it does die.

Anyway, please mail me (dsuvax!ghelmer@cs.utexas.edu) if you have any
data to contribute to my "survey" regarding length of uptime.
Please specify computer manufacturer, processor, load (low, moderate, high),
and mention anything that might be relevant.  Thanks.

-- 
Guy Helmer                             ...!bigtex!loft386!dsuvax!ghelmer
DSU Computing Services         ghelmer@dsuvax.uucp,  helmer@sdnet.bitnet
           Small is beautiful, but looks aren't everything...

sastdr@unx.sas.com (Thomas David Rivers) (07/17/90)

In article <1990Jul13.170652.1893@dsuvax.uucp> ghelmer@dsuvax.uucp (Guy Helmer) writes:
>I'm hoping to set up my old AT&T PC6300 to run as a 24-hr mail and
>public-domain server.  I have tried running Minix continuously on it
>without rebooting, but after two days it started acting wierd
>and after three it completely died (with very little system load!).
>
>How long has everyone else managed to keep Minix up, and on what
>systems?
>


  I had the same problems; which I eventually diagnosed as a memory leak.

  My system had a cron job running every minute (date>/tmp/date) just to
  make sure things were working, plus a uucp which cranked up about three
  times a day.

  After about two days, no job would be able to run (out of memory, 
  "Try again later.") - thus leading me to conclude the memory leak.

  This was present in 1.3, and 1.4a, on an XT clone.  I spent a little
  time trying to find the leak, but eventually gave up, and simply
  rebooted every morning.

  At the time, I posted a similar message describing the problem; but
  got no answer.  At this time, due to a move, my minix machine is down;
  so, I haven't researched or even upgraded to 1.5.10 (I am hoping the
  upgrade will remove the problem.)


		- Dave Rivers -

-- 
|  Time sharing: The use of many | - Dave Rivers - SAS Institute, Cary NC |
|   people by the computer.      |         sastdr@sas.com          (work) |
|				 |         rivers@ponds.UUCP       (home) |

root@minixug.hobby.nl (The Master) (07/18/90)

From article <1990Jul13.170652.1893@dsuvax.uucp>, by ghelmer@dsuvax.uucp (Guy Helmer):
> I'm hoping to set up my old AT&T PC6300 to run as a 24-hr mail and
> public-domain server.  I have tried running Minix continuously on it
> without rebooting, but after two days it started acting wierd
> and after three it completely died (with very little system load!).
Hmm...
 
> How long has everyone else managed to keep Minix up, and on what
> systems?
What about my machine, 'minixug' ?
This box (a 16MHz 286) runs NLMUG 1.5.10C-PC/286.  It has been running MINIX
since november '88.  I only switch to MS-DOS if I need to use WordPerfect
(5.10 NL) and my other machine ('uwalt') is in use.  In the past, I also
used to switch to DOS for formatting diskettes...

Once every month I make a total backup of my system, and, after that, I
reset the beast (note, that is _never_ turned off!).  My longest uptime
has been about 6-7 weeks at least.

I now have a third machine for all DOSy business, so I will have an uptime
of about 100% now.

In short: the machine only goes down when I tell it to.  No problems at
	  all (except for the processid-swap from 30000 to 1 !!).

> I suspect an 80[23]86 might keep Minix running longer,
> or point out problems better, than an 8086 or 68000, but I haven't
Hmm, not always.  The "uwalt" machine can run just as long, and it is
a 10MHz XT with V-20 cpu.  However, certain memory-related bugs (NULL
pointers!) will be catched by a 286 or 386, so the offending process
terminates instead of crashing the system.

> verified this.  I wish I could get kernel core dumps to find out
> what crashed the system when it does die.
Aaah, post-mortem dumps?  Should be possible with Bruce's db() installed!

> Anyway, please mail me (dsuvax!ghelmer@cs.utexas.edu) if you have any
> data to contribute to my "survey" regarding length of uptime.
> Please specify computer manufacturer, processor, load (low, moderate, high),
> and mention anything that might be relevant.  Thanks.

Sorry for _posting_ this, but I thought it would be a good remined for
all people running MINIX...

System:	FREECOM AT286/16	(16MHz 286, 0wt
				 1 to 16Mb (60ns), depending on the need
				 10 serial ports (COM1: COM2: and MUX)
				 2 modems: Discovery 1200-A and Trailblazer +
				 Hard Disk Seagate ST251-1 (40Mb, 28ms)
					   Seagate ST225-1 (20Mb, 40ms)

Software:			NLMUG Advanced MINIX 1.5.10C-PC/286

Load:				_Heavy_.
				This machine serves as a login base for all
				NLMUG members; it is the NLMUG Archive Server;
				it processes NEWS and MAIL (USENET, HobbyNET
				and the newly created MUGNET), it runs several
				Amoeba servers (File Server a la Netware,
				some Name Servers, ...); it runs some daemons
				that protect the system from bad users; and
				I use it as a software production system.

Uptime:				Always, unless I need it for MS-DOS work if
				my other machines are busy as well.

Note, that I do _not_ use "standard" MINIX; by now I know what needs to be
done to keep MINIX in the air...

Hope this helps,
	Fred van Kempen
	NLMUG System Manager |-)

PS:	Guy, please e-mail me.  I want to re-discuss the matter we talked
	about some months ago...

+-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-+
| MINIX User Group Holland	  UUCP: waltje@minixug.hobby.nl |
| c/o Fred van Kempen,		    or: hp4nl!minixug!waltje    |
| Hoefbladhof  27		                  		|
| 2215 DV  VOORHOUT						|
| The Netherlands	"A good programmer knows his Sources"	|
+-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-+

ghelmer@dsuvax.uucp (Guy Helmer) (07/19/90)

In <1990Jul13.170652.1893@dsuvax.uucp> ghelmer@dsuvax.uucp (Guy Helmer) writes:
>              I have tried running Minix continuously on it
>without rebooting, but after two days it started acting wierd
>and after three it completely died (with very little system load!).

>How long has everyone else managed to keep Minix up, and on what
>systems?

Thanks to everyone who has responded to my posting.  Every single
response (there were about 8 or 10) indicated that there is no
trouble with keeping MINIX up for extended periods.  One specific
case mentioned was the NLMUG machine, which is a mail/news machine
with very heavy daily use for news, mail, development, and whatever
else, and the only time it goes down is when it's taken down by
the admin.

Just to let everyone know that MINIX's reliability is much better 
than I expected :-).
-- 
Guy Helmer                             ...!bigtex!loft386!dsuvax!ghelmer
DSU Computing Services         ghelmer@dsuvax.uucp,  helmer@sdnet.bitnet
           Small is beautiful, but looks aren't everything...