SCO Login Problems
bernie
I am having a major headache with my SCO 5.0.6 login. When the server is
rebooted, I can login instantly with no problems at all. I can do this for a
week to 10days and then the login begins to slowly either time out or it
will take close to a minute to put in username and then the password. ANY
IDEAS how I can resolve this problem.
My machine is a P4, 1.0GB MEG, a raid controller with two 36GB SCSI Drive
to...@aplawrence.com
Possibly: http://www.aplawrence.com/SCOFAQ/scotec6.html#slowlogin
Is the performance otherwise acceptable?
See http://aplawrence.com/Unixart/slow.html
This doesn't sound like a network issue, but see
http://www.aplawrence.com/SCOFAQ/scotec4.html#telnetslow
for future reference
--
to...@aplawrence.com Unix/Linux/Mac OS X resources: http://aplawrence.com
Get paid for writing about tech: http://aplawrence.com/publish.html
Enrique Arredondo
"bernie" <ber...@on.aibn.com> wrote in message
news:oUF9b.5624$BT1.1...@news20.bellglobal.com...
I'm not sure if it's the same problem affecting 5.0.5 but I'm running this
command as root in order it to fix it :
tcbck
I've also another script that SCO created specially for me that will fix the
problem in case the above command doesn't work.
Peter McGill
Are you running a web server or something else?
I am running a box which just requires login during office hours.
What I do is schedule it to reboot every night.
If you know how to use vi, then do the following.
Log in as root.
Type: crontab -e
This will load root's crontab into the vi editor.
The add this line. (o is vi command to add new line.)
15 0 * * * /etc/shutdown -y -i6 -g00 > /dev/null 2>&1
Hit esc when done, then press :wq to save it.
Your server will now reboot at 15 mins past midnight.
Mine takes about 15 mins to restart, (waits on a few prompts for a
little while). Then it mails root an Auto Boot mail message.
On Tue, 16 Sep 2003 11:21:48 -0400, "bernie" <ber...@on.aibn.com>
wrote:
Bela Lubkin
> Do you need 100% up time?
> Are you running a web server or something else?
>
> I am running a box which just requires login during office hours.
> What I do is schedule it to reboot every night.
>
> If you know how to use vi, then do the following.
>
> Log in as root.
> Type: crontab -e
> This will load root's crontab into the vi editor.
> The add this line. (o is vi command to add new line.)
> 15 0 * * * /etc/shutdown -y -i6 -g00 > /dev/null 2>&1
> Hit esc when done, then press :wq to save it.
>
> Your server will now reboot at 15 mins past midnight.
> Mine takes about 15 mins to restart, (waits on a few prompts for a
> little while). Then it mails root an Auto Boot mail message.
Gee, I have a similar solution for my car. Every week or two it
develops this problem where it won't run. Something about "fuel", I
think. So I abandon it by the side of the road and buy a new one (they
always come with a full tank).
>Bela<
Bill Andersen
> Do you need 100% up time?
> Are you running a web server or something else?
>
> I am running a box which just requires login during office hours.
> What I do is schedule it to reboot every night.
Peter,
Not that this may not work as a temporary solution, but if it were
me, I'd rather figure out WHY the login gets slow and fix THAT problem.
I don't have an answer for "bernie" other than to keep looking for
the problem. This "fix" will work in the short run, but even if
you don't need 100% uptime, something isn't right on your box or
the login wouldn't continue to get slower. And what other problems
will get fixed when you figure out the cause of this one? Some you
may not even know about...
Peter, you haven't rubbed up against a Windows box lately have you?
Wanting to reboot every night looks like a symptom of the Windows Viri
<grin>
Bill
Peter McGill
wrote:
True Fixing the cause of the problem is best.
My system doesn't have a specific problem that requires it to reboot.
It's just been my experience, that every computer and every operating
system, wether unix, windows, or whatever needs to be rebooted every
once in a while. Even unix gets dead processes stuck in memory that
there is no other way to get rid of. Programs have memory leaks,
etc... If you don't have a regular reboot schedule then eventually the
problems stack up and cause noticable side effects, even bringing down
the whole server. True unix is better at managing these problems than
other operating systems, but it is not immune.
Since our severs don't need to be up all day, I just get ahead of the
problem and reboot frequently. Simple but effective.
Peter
Bill Vermillion
>On Tue, 7 Oct 2003 20:07:43 GMT, Bill Andersen <bi...@mwdental.com>
>wrote:
>
>>Peter wrote:
>>> Do you need 100% up time?
>>> Are you running a web server or something else?
>>> I am running a box which just requires login during office
>>> hours. What I do is schedule it to reboot every night.
>>Peter,
>>Not that this may not work as a temporary solution, but if it
>>were me, I'd rather figure out WHY the login gets slow and fix
>>THAT problem. I don't have an answer for "bernie" other than to
>>keep looking for the problem. This "fix" will work in the short
>>run, but even if you don't need 100% uptime, something isn't
>>right on your box or the login wouldn't continue to get slower.
>>And what other problems will get fixed when you figure out the
>>cause of this one? Some you may not even know about...
>>Peter, you haven't rubbed up against a Windows box lately have
>>you? Wanting to reboot every night looks like a symptom of the
>>Windows Viri <grin>
>
>True Fixing the cause of the problem is best.
>My system doesn't have a specific problem that requires it to
>reboot.
>It's just been my experience, that every computer and every
>operating system, wether unix, windows, or whatever needs to be
>rebooted every once in a while. Even unix gets dead processes
>stuck in memory that there is no other way to get rid of.
>Programs have memory leaks, etc... If you don't have a regular
>reboot schedule then eventually the problems stack up and cause
>noticable side effects, even bringing down the whole server.
>True unix is better at managing these problems than other
>operating systems, but it is not immune.
I've never noticed that on the servers I run - they just keep
running and running.
I do make it a point to do a complete OS upgrade every two years
however and that requires a reboot. Around the first of the year I
rebuilt the entire OS on the webserver and it's been up for 245
days now - while before I restarted it [actually moved to a new
machine that was indentical except for a much larger hard drive]
it was at about 780 days. It just never falters, hicups or slows
down. It peaks in the spring running about 80-90K hits/day.
Comes up on the top of the Google sites because of a stong domain name,
so that helps the traffic.
The mail machine has shown a bit of slowness in bursts lately but
that because of the huge amount of spam coming in - and during
those peaks it's getting about 10/second - and rejecting about 75%
of them :-)
A machine I remotely maintain in D.C. just was rebooted two week
ago after running for 750+days. They lost power during the big
storm and it is not in a colo.
I've stopped having clients reboot their SCO systems since the
lbolt problem was fixed. I did notice a flaw in the time on an
OSR 5.0.5 - in that about after 14 months the uptime in w/who went
away. Totally blank.
In the past 5+ years the OSes have become very stable from my POV.
Maybe you have hardware problems.
I've never had one of these machine just stop working. Fairly
plain machines - the servers are running iNTEL mobos with some
Tyans - the OSR5 are most often ASUS.
>Since our severs don't need to be up all day, I just get ahead
>of the problem and reboot frequently. Simple but effective.
Well mine need to be up all the time but never need to be rebooted.
The one exception was a dedicated server that unexplicably lost
one of the NIC ports on the mobo. So that one required a hands
on visit and a reboot.
Bill
--
Bill Vermillion - bv @ wjv . com
Bill Andersen
> True Fixing the cause of the problem is best.
>
> My system doesn't have a specific problem that requires it to reboot.
>
> It's just been my experience, that every computer and every operating
> system, wether unix, windows, or whatever needs to be rebooted every
> once in a while. Even unix gets dead processes stuck in memory that
> there is no other way to get rid of. Programs have memory leaks,
> etc... If you don't have a regular reboot schedule then eventually the
> problems stack up and cause noticable side effects, even bringing down
> the whole server. True unix is better at managing these problems than
> other operating systems, but it is not immune.
>
> Since our severs don't need to be up all day, I just get ahead of the
> problem and reboot frequently. Simple but effective.
I don't guess it will hurt anything. I just don't really see the
need on a properly configured OS. I've not found the "memory leaks"
or dead processes to be a problem. Granted, I've only got about 30 users
running a specific accounting/distribution package. Not too many other
programs running. However, I've got a couple of web servers, mail server
and FTP machines on Unix. I don't reboot them unless I am doing a hardware
upgrade or something that absolutely requires a reboot/shutdown.
No real long uptimes at this point (172 days), but the only reason the
machines were shutdown at that time was because we physically moved
3 machines from one room to another new "dedicated" room. I found it
hard to unplug them without causing some downtime <grin>! Previous
uptime was most likely upwards of a year, but I didn't check before
we moved them.
In any event, I think the whole concept of restarting on a regular
frequency has become an accepted way of thinking because of Microsoft.
I interviewed a recent college graduate. He knew absolutely nothing
other than Windows. When describing his responsibilities with a
previous employer, he indicated he did "Preventive Maintenance" on
their Win2000 server farm. I asked him what that included and he
said "Install security patches, update application software and reboot
each server daily at 5:30am to 'clear memory'" His reason for
looking for another job was because he didn't like getting to work
that early!!!
Why, oh why, has this just become acceptable practice?
Bill
Peter McGill
wrote:
>Peter wrote:
Microsoft why else?
Doesn't everyone hate them, everyone who knows anything anyway.
I'm trying to migrate all our servers to Linux, We're about 40%
Win2000 now. Ew. But of course the people with the money who make the
decisions won't listen to us educated folk who actually know what
we're doing.
I've only had a SCO server run for about 6 months, then I had to
reboot it, the guy who installed it, did just that, installed and
left, setup no maintenance of any kind, reboots asside, you still need
regulary scheduled maintenance.
Anyway I will shutup about reboots.
Have we fixed bernie's problem yet, lets not forget about him.
Peter