Unstable ODT3 server... At wit's end

Unstable ODT3 server... At wit's end

Post by My » Tue, 23 May 1995 04:00:00



Hello.  We're having some problems with our SCO ODT3 server/development
machine that can only be described as bizarre.  Frankly, we're
considering burning the hard drive, using the tapes as confetti and
calling an exorcist.

It all started with trying to get the machine to make a PPP connection.
During this experience (which all SCO ODT3 owners can tell you is a
!fun thing to spend weeks on) we accidentally fried our Computone
serial multiplexer (despite the APC backup/surge protector, a big
expensive one (thank goodness for their $25000 guarantee) (yes, too many
parentheses)).  After getting a replacement, things got really weird.

First of all, we were unable to configure a new modem on the Computone.
The tech support with them said that it must be a problem with the OS
since the current modems and terminals on the box still worked fine.
Heck, we wanted to run PPP through the serial lines, but gave up on
that after crashing the server by disabling/turning off the modems we
were trying to add.  So we settled on one of the two physical ports.

Eventually, we did get PPP working on the SCO (thank you for your help,

xmission.com).  Slowly, we added a name server, then SMTP to supplement
our UUCP connection.  We even got the other PC in the office to connect to
the net using the SCO machine and its PPP link as a gateway using the
installed Western Digital Ethernet cards (on both machines) and Trumpet
Winsock.

But then we got*y, and upgraded the PC in the office with Windows
for Workgroups, added the free MS TCP/IP stacks, and added a GVC
Ethernet card to another PC in the office, configuring it too with WfWG
and MS TCP/IP.

The result?  The PCs can talk to each other and the SCO machine, but
cannot communicate past the SCO machine.  They can ping each other and
the SCO, but a traceroute to even the other end of the PPP line causes
a Network Unreachable error (without even acknowledging the SCO machine).
We know our nameserver was functioning, because it was resolving the
IP addresses properly (even starting a PPP connection to get them).
Seems like a routing problem, right?  We've tried all sorts of solutions.
Even the handy TCP-IP Network Administration book from O'Reilly isn't
a help.

And now it appears that the Adaptec 1540 controller is toasted.  Our
backups have at least a couple of "Out of sync" errors on them every
night, even if it's a new tape or a new SCSI tape drive.  Of course,
the wonderful SCO documentation we've got doesn't document this error
anywhere.  And it almost seems like the SCSI CDROM is giving us errors
too, saying it can't install all the files in a package while using
custom.

Heck, the entire server crashed on me early last week when I did
an innocent "tcp stop" for the first time since a reboot three days
before.

Is it time to call in professional help?  Should we inscribe a
pentagram around our SCO box?  Sacrifice a fresh copy of Linux to it?
Re-install everything from scratch?  (Remember, our backups are not
too reliable).  We've tried isolating our problems and dealing with
them one at a time, but it appears that the system is unstable and
we are getting very frustrated.

Ah yes, and one last thing.  Money is extremely tight, so we can't
afford to throw dollars after this.  Us technical people have to
perform a miracle.  So we'd appreciate any help you could give us.

Thanks a million for listening (and it's stress-relieving to tell).

Michael Smit, The Other Technical Guy (compared to the one who's
been there for a year and a half already)

--
]\\ `/ ]< (Michael C. Smit) /\ Delirium: "There must be a word for it... The

      Welcome to Utah       \/ Is there a word?"  Dream: "Change."

 
 
 

Unstable ODT3 server... At wit's end

Post by D. Chris Daniel » Wed, 24 May 1995 04:00:00


~
~It all started with trying to get the machine to make a PPP connection.
~During this experience (which all SCO ODT3 owners can tell you is a
~!fun thing to spend weeks on) we accidentally fried our Computone
~serial multiplexer (despite the APC backup/surge protector, a big
~expensive one (thank goodness for their $25000 guarantee) (yes, too many
~parentheses)).  After getting a replacement, things got really weird.
~

PPP realy isn't that bad to configure...  Unless you have not installed one of
the three sls patches :-)

~
~The result?  The PCs can talk to each other and the SCO machine, but
~cannot communicate past the SCO machine.  They can ping each other and
~the SCO, but a traceroute to even the other end of the PPP line causes
~a Network Unreachable error (without even acknowledging the SCO machine).
~We know our nameserver was functioning, because it was resolving the
~IP addresses properly (even starting a PPP connection to get them).
~Seems like a routing problem, right?  We've tried all sorts of solutions.
~Even the handy TCP-IP Network Administration book from O'Reilly isn't
~a help.

So you are saying that you can not get out the ppp connection from the pc's??  
Did you set the default gateway for the PC to be the UNIX host??  And I am
assuming that you are using some windows based telnet to connect to the UNIX box
and trying a traceroute from there.  More info please.

I connect from home to our OSE3.0 box with trumpet using PPP and then get out on
the net via our dedicated ppp connection.  This was all a no brainer.

        **chris

~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
D. Chris Daniels
Consulting/Technical Sales      SCO ACE


 
 
 

Unstable ODT3 server... At wit's end

Post by Bill Vermilli » Tue, 30 May 1995 04:00:00




>Hello.  We're having some problems with our SCO ODT3 server/development
>machine that can only be described as bizarre.  Frankly, we're
>considering burning the hard drive, using the tapes as confetti and
>calling an exorcist.
>It all started with trying to get the machine to make a PPP connection.
>During this experience (which all SCO ODT3 owners can tell you is a
>!fun thing to spend weeks on) we accidentally fried our Computone

                              ^^^^^^^^^^^^^^^^^^^^^^

Quote:>serial multiplexer (despite the APC backup/surge protector, a big
>expensive one (thank goodness for their $25000 guarantee) (yes, too many
>parentheses)).  After getting a replacement, things got really weird.

How did you accidentally fry this.  Knowing what caused it
might help.  I've seen some hardware failures make other pieces
of hardware flaky - eg work most of the time but unexptected
failured.

Quote:>First of all, we were unable to configure a new modem on the Computone.

What errors do you get?

Quote:>But then we got*y, and upgraded the PC in the office with Windows
>for Workgroups, added the free MS TCP/IP stacks, and added a GVC
>Ethernet card to another PC in the office, configuring it too with WfWG
>and MS TCP/IP.
>The result?  The PCs can talk to each other and the SCO machine, but
>cannot communicate past the SCO machine.  They can ping each other and
>the SCO, but a traceroute to even the other end of the PPP line causes
>a Network Unreachable error (without even acknowledging the SCO machine).

Did the traceroute work properly before you did this?

Quote:>And now it appears that the Adaptec 1540 controller is toasted.

See my previous line about one hw failure causing others.  OTOH
about a year ago a vendor friend of mine was getting 30%
failure rates on the 154x series.   Appears to have been a 2 or
3 month problem since cured.  Showed up mostly on CD-ROMS

Quote:>Our
>backups have at least a couple of "Out of sync" errors on them every
>night, even if it's a new tape or a new SCSI tape drive.

Your first problem when you 'accidentally fried' the Computone
could be involved here.      

Quote:>Heck, the entire server crashed on me early last week when I did
>an innocent "tcp stop" for the first time since a reboot three days
>before.

I think you have some hardware problem that is intermittent.  I
had one SCO box that would crash for no apparent reason - and
the error messages when refering to the error message area in
the manual said this could be caused by hardware or software.
No one I talked to had seen this message.  And that included
some APCs and support people who have dealt with 100's of
machines.   Changing the serial board fixed that one.  Maybe
you have infant mortality on the new Computone board.

Quote:>Is it time to call in professional help?  Should we inscribe a
>pentagram around our SCO box?  Sacrifice a fresh copy of Linux to it?
>Re-install everything from scratch?  (Remember, our backups are not
>too reliable).  We've tried isolating our problems and dealing with
>them one at a time, but it appears that the system is unstable and
>we are getting very frustrated.

Sounds like a hardware problem to me.   It could be as low
level as the motherboard.  Maybe you can 'borrow' another
system to test that theory.

--

 
 
 

Unstable ODT3 server... At wit's end

Post by Bma b » Thu, 08 Jun 1995 04:00:00


You might also try putting the boards one by one in another server and
seeing if they fail.

Another thing to try might be re-loading your O/S from scratch, especially
if you are getting bogus numbers from sar on resource useage.
Brian Campbell
System Administrator
B&M Associates

 
 
 

1. PPP configuration problem - (I'm at my wit's end)

Hello,  I'm brand new to the Linux world.  I've read every related
document that came with my distribution and still can't figure out
what's wrong.

I installed Linux and X11 from a new Infomagic distribution.
Everything went very smoothly and configuration was a breeze, except
for a strange problem with PPP.

Using dip, I connect to my ppp server via 14400 modem, send my
username, password, and issue a ppp and end with 'mode PPP'.  All of
the IPs are dynamic (that is the IP assinged to my machine and the IP
of the machine I'm connected to) This works fine - I can telnet,
finger, ftp, ping, etc.  When I use ifconfig and route to look at the
current network status everything looks OK - my IP and the P-t-P IP
look OK.

HOWEVER, I can't get Netscape 2.02 working correctly.  If I give it a
host name it gets the IP and says "Connect:  Host conntacted, waiting
for reply" or something to that effect, but I never get any reply -
its like Netscape can find the host but the host can't send anything
back.  

Is this a routing problem?  How do I configure it so that programs
other than telnet, ftp, etc. can work.

I've never done any network configuration so I'm not really sure how
to fix this problem.

jeff lee.  

2. How to use Wget for mozilla downloads?

3. FPC Setup "Free Pascal Compiler, I am at wit's end"

4. PGI compiler problem in RH6.0

5. ?: how is the loopback created; at wit's end

6. Packet Padding

7. NIS problems (at wit's end)

8. telnetd, ftpd failing: can't login remotely after apache install

9. Help, I'm at my wits end over PPP

10. At wit's end with RealTek 8139c

11. At wit's end...

12. Wit's end question -- files go to print spooler and are never printed

13. At wit's end: Apache+PHP4: DSO Garbled message