Strange Characters during boot up

[Robert Gamble]

Wow! never seen that one before ...

Maybe something in one of the /sbin/init.d scripts is changing $TERM settings for output ?

Good Luck!

[Clemens van Everdingen]

Hi,

Also never seen before.
Using a 712 myself.
Ever need a manual for this thing I have one
http://cmve.dtdns.net select manuals.

I guess it has something to do with the LANG setting ??

C.

[Kelli Ward]

Hi Guys,

I have thought about that, but it's interesting that it changed back at login.

I haven't investigated /sbin/ files, (Notice it starts while outputing /sbin)but am planning on it. The times I've seen it in the past, I've ended up doing a quick rebuild on the system (due to time constraints), but I've seen it enough times where if there is a non-rebuild solution, I'd love to know it.

I'll look at the manuals
Thanks,
Kel

[Jeff Schussele]

Hi Kelli,

Ditto - that a new one.

Possibly trying to cat/echo a binary? Check the issue, motd, etc.

HTH,
Jeff

[Kelli Ward]

Hi Clemens,
I don't see a manuals hyperlink on your page.
Do I type it at the search prompt?
Kel

[Clemens van Everdingen]

Kelli,

Just go to links and then select manuals from the drop down box.

C.

[Kelli Ward]

Hi Clemens,
I found the manual thanks. I didn't notice the pop-down menu.

Open to all - Still need help on this problem if anyone has any ideas. I inadvertantly fibbed on the system type. The model is actaully a 743i/B not a 712 as previously indicated. There are three workstations where the ailing unit is, so I mixed up my models.

Any ideas are greatly appreciated.

Thanks,
Kel

[David Bell_1]

Kelli,

If not already installed, you may want to try the PHNE_21872 serial, mux, pty cumulative patch. There are some things associated to XON/XOFF and RTS/CTS flow control within this patch.

HTH,

Dave

[Kelli Ward]

Hi
Thanks for the patch.
It didn't help, unfortunately, but I appreciate the effort.

Any other brave souls willing to help?

Thank,
Kel

[Sanjay_6]

Hi Kelli,

Check the terminal setting on the console. Also before booting, switch off the console and then switch it on. Then start the system boot. The starange chanracters could be becuase of someone earlier logged in through the console and having not specified his / her termtype and hp. Instead could have used some other terminal type, say vt220 for example. The display is therefore overlapping with the previous display.

Just power off and power on the terminal and see if those strange characters are still displayed. Once the login screen comes on the console and you login as root, you terminal type is set to hp and hence you don'e see any overlapping of characters.

Hope this helps.

Regds

[David Bell_1]

Kelli,

What kind of terminal is attached to the serial port?

If it's a 700/9X, Sanjay may have a point. To reset to default values including the term type;

Power off the monitor.
Wait for about 10 seconds.
Power the monitor back on while holding down the 'D' key. This will set the terminal back to the default values.

HTH,

Dave

[Kelli Ward]

Hi,
Sanjay - I attempted your suggestions with no luck.

I am actually using a Digi 8 port terminal emulator on a 712 (hence the original model mix up)

It is running through serial. I reset the terminal server and the 712, as well as the 743i with the issue like I said with no luck. I have noticed nothing in the startup files. (that caught my eye, but I may be/am missing something)
All other emulated workstations (2) are running fine.
Another tidbit. I had mentioned seeing this in the past. It was on C110 systems (if memory serves, it was a couple of years ago) all running console mode not serial.
About the only thing I haven't attempted is to change the actual connections and try a different port on the emulator.

Note: There have never, to my knowledge, been any overlapping terminals, which can cause strange behavior like this. ps on the 712 only shows one process open for this port.
Most of the overlapping data you are seeing is actually the boot sequence where it starts a page worth of rc.# scripts clears the screen runs more rc.# scripts, etc. until it finally comes up with the login prompt when the output returns to a readable state. And this unreadable character output is not entered into the /etc/rc.log which looks perfectly fine and readable and no error logged.
It's such an odd behavior that I have to believe it is some sort of environment setting, but I can't find it.
kbdlang is PS2_DIN_US_ENGLISH
I have also checked all system logs and noticed nothing unusual.

I'm almost at the point of a rebuild, but just have to think there is an easier way.

help?... ;)

Kel

[Jeff Schussele]

Kelli,

I agree it looks like its garbaging the rc.x ouput. But before working thru these via /sbin/checkrc script it works thru the /etc/inittab using /sbin/ioinitrc script & /etc/ioconfig as its config then after checkrc it goes thru /sbin/rc - so definitely check all of these for corruption. Compare to known good systems.

Also you may want to consider that the driver for the Digiport 8 has been corrupted - maybe reloading it will cure the problem.

I know - I'm like you I hate to reload unless there's NO other way.

My 2 cents,
Jeff

[Kelli Ward]

Hi Jeff,

I actually don't mind a good reload, if I know it is the best, fastest way to acheive my goals.
Although I've spent more time on this than a reload, I 've been going for the grater good approach. That and satisfy my curiosity.

Except for checkrc, I've been through all mentioned files. I'm going there next.

Question -
1.) What sends the Memory Information to terminal right before the text garbles?

2.) Could this be a STREAMS problem?
(admittedly strams-ux is an enigma to me)
I haven't taken much time to study it and only barely know what it does.

Thanks,
Kel

[Kelli Ward]

Jeff!
You got me!
I jumped on "checkrc" like a hungry fish to a hooked worm.
Yes I have checked bcheckrc - nope.
Reload is starting to look more and more attractive.
Even if it was accidental, thanks for the levity Jeff.
I know, you were going for that whole pre-April Fool's thing weren't you? ;)

Thank,
Kel

[Jeff Schussele]

Well the order as I understand it is:

1) PDC (processor dependent code)
2) ISL (initial system loader)
3) Secondary Loader (hpux)
4) kernel (/stand/vmunix)
5) swapper (pid 0)
6) /sbin/pre_init_rc (runs fsck & other)
7) init (pid 1)


So my guess would be either vmunix or the pre_init_rc script.

And no I doubt it's Streams - it's got to be something involved with ioinitrc as that immediately follws the mem check.

Rgds,
Jeff

[Kelli Ward]

and streams-ux is even more of an enigma than strams-ux.

Man, I need my Easter supper.
Is this day over yet? ;)
Kel

[Kelli Ward]

Hi,
I've been over pre_init_rc and booted vmunix.prev
I hadn't considered swap, but I can check.
I'm also going to try a new terminal, I haven't quite gotten back there yet, been doing most of my work through telnet.
Thanks,
Kel

[Kelli Ward]

Hi all,
New terminal was a bust.

Have a great weekend all!
We can do it again on the other side.

Kel

[Sanjay_6]

Hi Kelli,

Suspect the strange characters appearing on your screen has something to do with the terminal emular you are using. Connect a standard console directly to the system and see if the strange characters still appaear on the screen. Also check the serial port configuration. Set the port configuration to 9600, 8 bits, no parity, emulation type hp and try.

Hope this helps.

Regds

[Brian butler]

could you not also check by remotely logging onto this machine, weather or not this is a console problem...

[Kelli Ward]

Hi all,
I've tried an alternate terminal, with no luck.
The terminal is set up 9600, 8 bit no parity and the strange characters revert back to a normal output at login, so (unfortunately) a remote login doesn't tell me anything.
Thanks,
Kel

[Jeff Schussele]

Hi Kel,

Ok - I'm ready to give it another go.

1) Have you tried Sanjay's advice to hook a console direct to cons port? What was result?
2) I'm not sure I know exactly what/how this Digi 8 is/works. Is it a net device or serial. Are there inittab entries for ea port? Have you checked the getty params for the port? Hidden chars? switched ids? Tried removing the -h (if there) from the getty command to force a hangup before speed set? Have you checked the /etc/gettydefs file with the getty -c command?
3) Unprintable filenames in the rc.# dir(s)?
4) Any README files in the rc.x dir(s) w/exeute perm?
5) Relinked a kernel (w/no changes) just to verify that /etc/ioconfig & the kernel are in sync?

I suspect, as Sanjay, that it's the port off the Digi getting goofed (that'a a highly technical term :) during ioinit.
But let's not rule out the other stuff.

HTH,
Jeff

[Kelli Ward]

Hi,
I don't think this problem is on the system with the Digi attached.

Let me backup a bit.

The Digi 1008 (not 8, sorry) port server is a device that attaches to a system through the narrow scsi bus, in this case the 712. It has 8 db25 connectors on it that once the drivers are installed on the 712, can emulate 8 serial terminals. You can connect up to 8 workstations to it and have access to each one through an hpterm or dtterm using cu.

The issue I'm having is with the 743i which I removed from the digi and connected to a seperate terminal this morning, still gave me the same garbled looking output. The issue seems to be in the boot up sequence of the 743i drive, but I can't seem to find where.

The Digi server is using /etc/uucp/Devices not /etc/inittab for it's cu setup and has no configuration error for it's device file.
I have not looked into the README suggestion, but will. I have rebuilt both ioconfig files.

I was hoping for a solution that didn't require a full rebuild, but it's looking more and more like that may be my solution.

Thanks,
Kel