Mailing List Archive

>Can someone tell me if Solaris 8 and 9 still goes into the boot monitor
>when a break is sent across the serial line?

Yes, it does. And what a major pain in the ass that is.

This "feature" has been discussed on this list before, but I don't recall
seeing a viable solution.

>Chris

John R. Jackson, Technical Software Specialist, ITaP/RCS, jrj@purdue.edu

Hoe would one get into the boot monitor from a tty-rsc console. This is
a console over ethernet. Sending a break via conserver would be of no
use.


On Mon, 2003-03-03 at 09:38, Sigurd Mytting wrote:
> [Kjell Andresen]
>
> | On Mon, 3 Mar 2003, cfowler wrote:
> || Can someone tell me if Solaris 8 and 9 still goes into the boot monitor
> || when a break is sent across the serial line?
>
> | I guess so, but you may change the break sequence on your Solaris
> | host(s).
>
> It does. You can change the break-sequence by altering KEYBOARD_ABORT
> in /etc/default/kbd; by setting it to 'alternate' the break-sequence
> is set to '<RETURN><TIDLE><CONTROL B>'.
>
>
>
> -Sigurd
> --
> Sigurd Mytting, <URL:http://www.usit.uio.no/> Section for Operations
> Center for Information Technology Services, University of Oslo, Norway
--
"The Law of Leaky Abstractions"
There is a time where abstractions lead to the inablity to
fix problems that leak through the abstraction.
http://www.joelonsoftware.com/articles/LeakyAbstractions.html

On 3 Mar 2003, cfowler wrote:

> Hoe would one get into the boot monitor from a tty-rsc console. This is
> a console over ethernet. Sending a break via conserver would be of no
> use.

one of the commands from the rsc> prompt is 'break'.
(but you can still use serial A to send break in the normal matter too).
If you want to use RSC as the machine console, you have to set the
eeprom to tell it to use RSC as the system console. We just use
regular console as console and RSC for emergency supplement so
I don't remember the exact syntax, but a google search should turn
it up.

>
>
> On Mon, 2003-03-03 at 09:38, Sigurd Mytting wrote:
> > [Kjell Andresen]
> >
> > | On Mon, 3 Mar 2003, cfowler wrote:
> > || Can someone tell me if Solaris 8 and 9 still goes into the boot monitor
> > || when a break is sent across the serial line?
> >
> > | I guess so, but you may change the break sequence on your Solaris
> > | host(s).
> >
> > It does. You can change the break-sequence by altering KEYBOARD_ABORT
> > in /etc/default/kbd; by setting it to 'alternate' the break-sequence
> > is set to '<RETURN><TIDLE><CONTROL B>'.
> >
> >
> >
> > -Sigurd
> > --
> > Sigurd Mytting, <URL:http://www.usit.uio.no/> Section for Operations
> > Center for Information Technology Services, University of Oslo, Norway
> --
> "The Law of Leaky Abstractions"
> There is a time where abstractions lead to the inablity to
> fix problems that leak through the abstraction.
> http://www.joelonsoftware.com/articles/LeakyAbstractions.html
>
> _______________________________________________
> users mailing list
> users@conserver.com
> https://www.conserver.com/mailman/listinfo/users
>

The command is:

ok setenv input-device rsc
ok setenv output-device rsc
ok diag-output-to rsc

These older commands will not work in redirecting the console to RSC.
The following must be used in order for the redirection to take effect:

ok setenv diag-out-console true
ok input-device rsc-console
ok output-device rsc-console

These commands will take effect after the next server reset.
If you are wanting to redirect the console server back to TTYA , you can
do the following:

ok setenv diag-out-console false
ok setenv input-device keyboard
ok setenv output-device screen


> Hoe would one get into the boot monitor from a tty-rsc console. This is
> a console over ethernet. Sending a break via conserver would be of no
> use.

one of the commands from the rsc> prompt is 'break'.
(but you can still use serial A to send break in the normal matter too).
If you want to use RSC as the machine console, you have to set the
eeprom to tell it to use RSC as the system console. We just use
regular console as console and RSC for emergency supplement so
I don't remember the exact syntax, but a google search should turn
it up.

The only solution which I have found that gives satisfactory
results is the Nudata(TM) Non Aborting Serial Port Adaptor. This
is a little black box that cost about $90. It has smarts onboard
and allows the BREAKs to be sent from a terminal or terminal
server port. But it monitors the DTR line, if it drops then the
box blocks all input, thus if a terminal server sends breaks at
power down they are blocked when DTR drops first. There is a
timing issue here but it seems to work, I suspect they have a
transmission delay built in to catch the fault before the BREAK is
sent but I don't know this for sure.

The beauty of it is that we can still use BREAK to halt the system
if we want. If the terminal server looses power all our Sun's
DON'T halt. It would also cover off any problems of the line being
cut and looking like a break as long as DTR is cut at the same
time.

The only down side is the price tag.

"John R. Jackson" wrote:
>
> >Can someone tell me if Solaris 8 and 9 still goes into the boot monitor
> >when a break is sent across the serial line?
>
> Yes, it does. And what a major pain in the ass that is.
>
> This "feature" has been discussed on this list before, but I don't recall
> seeing a viable solution.
>
> >Chris
>
> John R. Jackson, Technical Software Specialist, ITaP/RCS, jrj@purdue.edu
> _______________________________________________
> users mailing list
> users@conserver.com
> https://www.conserver.com/mailman/listinfo/users

--
Mike Daigle \ | |
Network Operations Project Supervisor \| /\ |~ |/
Computing and Network Services | \/ | |\
Room 030 Steacie Science Library UNIVERSITY
Email: mdaigle@yorku.ca 4700 Keele St. Toronto
Phone: 736-2100 ext.22730 Ontario, Canada M3J 1P3

For thousands of years mankind lived just like the animals;
Then something happened that unleashed the power of his
imagination;
He learned to talk.
-- Pink Floyd

On Tue, 4 Mar 2003, Mike Daigle wrote:

> The only solution which I have found that gives satisfactory
> results is the Nudata(TM) Non Aborting Serial Port Adaptor. This
> is a little black box that cost about $90. It has smarts onboard
> and allows the BREAKs to be sent from a terminal or terminal
> server port. But it monitors the DTR line, if it drops then the
> box blocks all input, thus if a terminal server sends breaks at
> power down they are blocked when DTR drops first. There is a
> timing issue here but it seems to work, I suspect they have a
> transmission delay built in to catch the fault before the BREAK is
> sent but I don't know this for sure.
>
> The beauty of it is that we can still use BREAK to halt the system
> if we want. If the terminal server looses power all our Sun's
> DON'T halt. It would also cover off any problems of the line being
> cut and looking like a break as long as DTR is cut at the same
> time.
>
> The only down side is the price tag.

To summarize what some others have said (and add a couple more)

The Digi, old GMP boxes (don't think they are still around), cisco
26xx and 36xx series Async cards, Lantronix (SCS-16xx, SCS-32xx), Bay
Networks (Annex), AlterPath ACS16, and others: all work as they should.

See http://www.conserver.com/consoles/breakinfo.html for
a rather exhaustive list of such boxes that work fine with breaks
(even when power cycled)


Doug

Solaris now has the file /etc/default/kbd file where you can turn off
the break from the console server.

"KEYBOARD_ABORT affects the default behavior of the keyboard abort
sequence, see kbd(1) for details. The default value is "enable". The
optional values are "disable" or "alternate". Any other value is ignored.
If you choose "alternate" it will affect the serial console drivers ONLY.
The keyboard BREAK (sequence and plug/unplug) won't be affected by this.
If "alternate" is in effect any protocol (PPP, SLIP... etc) should not be
run over the serial console port."

You can make your changes and reboot the box, or make the changes and
then run "kbd -a alternate" from command line.

This is support by default for Solaris 8/9, for 2.6 & 7 you need to
install the patches; 105924-10 and 107589-02.

http://docs.sun.com/db/doc/835-8001/6rutvdvja?a=view

http://www.sunmanagers.org/pipermail/summaries/2001-August/001235.html

Steve

Mike Daigle wrote:

>The only solution which I have found that gives satisfactory
>results is the Nudata(TM) Non Aborting Serial Port Adaptor. This
>is a little black box that cost about $90. It has smarts onboard
>and allows the BREAKs to be sent from a terminal or terminal
>server port. But it monitors the DTR line, if it drops then the
>box blocks all input, thus if a terminal server sends breaks at
>power down they are blocked when DTR drops first. There is a
>timing issue here but it seems to work, I suspect they have a
>transmission delay built in to catch the fault before the BREAK is
>sent but I don't know this for sure.
>
>The beauty of it is that we can still use BREAK to halt the system
>if we want. If the terminal server looses power all our Sun's
>DON'T halt. It would also cover off any problems of the line being
>cut and looking like a break as long as DTR is cut at the same
>time.
>
>The only down side is the price tag.
>
>"John R. Jackson" wrote:
>
>
>>>Can someone tell me if Solaris 8 and 9 still goes into the boot monitor
>>>when a break is sent across the serial line?
>>>
>>>
>>Yes, it does. And what a major pain in the ass that is.
>>
>>This "feature" has been discussed on this list before, but I don't recall
>>seeing a viable solution.
>>
>>
>>
>>>Chris
>>>
>>>
>>John R. Jackson, Technical Software Specialist, ITaP/RCS, jrj@purdue.edu
>>_______________________________________________
>>users mailing list
>>users@conserver.com
>>https://www.conserver.com/mailman/listinfo/users
>>
>>
>
>
>

Digi makes a line of terminal servers that have this feature built in.
They are reasonably priced as well. I use the Portserver II, but
bought them some time ago, other recently offered products may be a
better fit.

http://www.digi.com/products/index.jsp
http://www.digi.com/products/terminal%20servers/portServerII.jsp

Mike Daigle wrote:

>The only solution which I have found that gives satisfactory
>results is the Nudata(TM) Non Aborting Serial Port Adaptor. This
>is a little black box that cost about $90. It has smarts onboard
>and allows the BREAKs to be sent from a terminal or terminal
>server port. But it monitors the DTR line, if it drops then the
>box blocks all input, thus if a terminal server sends breaks at
>power down they are blocked when DTR drops first. There is a
>timing issue here but it seems to work, I suspect they have a
>transmission delay built in to catch the fault before the BREAK is
>sent but I don't know this for sure.
>
>The beauty of it is that we can still use BREAK to halt the system
>if we want. If the terminal server looses power all our Sun's
>DON'T halt. It would also cover off any problems of the line being
>cut and looking like a break as long as DTR is cut at the same
>time.
>
>The only down side is the price tag.
>
>"John R. Jackson" wrote:
>
>
>>>Can someone tell me if Solaris 8 and 9 still goes into the boot monitor
>>>when a break is sent across the serial line?
>>>
>>>
>>Yes, it does. And what a major pain in the ass that is.
>>
>>This "feature" has been discussed on this list before, but I don't recall
>>seeing a viable solution.
>>
>>
>>
>>>Chris
>>>
>>>
>>John R. Jackson, Technical Software Specialist, ITaP/RCS, jrj@purdue.edu
>>_______________________________________________
>>users mailing list
>>users@conserver.com
>>https://www.conserver.com/mailman/listinfo/users
>>
>>
>
>
>

[. On Tuesday, March 4, 2003 at 09:12:44 (-0500), Mike Daigle wrote: ]
> Subject: Re: Solarus 8/9 Sparc Questions
>
> The only solution which I have found that gives satisfactory
> results is the Nudata(TM) Non Aborting Serial Port Adaptor. This
> is a little black box that cost about $90. It has smarts onboard
> and allows the BREAKs to be sent from a terminal or terminal
> server port. But it monitors the DTR line, if it drops then the
> box blocks all input, thus if a terminal server sends breaks at
> power down they are blocked when DTR drops first. There is a
> timing issue here but it seems to work, I suspect they have a
> transmission delay built in to catch the fault before the BREAK is
> sent but I don't know this for sure.
>
> The beauty of it is that we can still use BREAK to halt the system
> if we want. If the terminal server looses power all our Sun's
> DON'T halt. It would also cover off any problems of the line being
> cut and looking like a break as long as DTR is cut at the same
> time.
>
> The only down side is the price tag.

Indeed.

Time to (again?) post the link to circuit diagram of what I believe is
an equivalent device.

The diagram was originally posted in sun-spots v5n45 by Malcolm Harper,
back before the WWW and most copies are in compressed archives and thus
not indexed by the likes of Google. A copy can now be found at this
location:

http://sunsite.kth.se/sun/sun-spots/v5n45.Z

The relevant post has been re-copied to at least these locations
subsequently:

http://sunsite.bilkent.edu.tr/pub/sun-info/sunspots/v9n20
http://mail-index.netbsd.org/port-sparc/2002/05/16/0000.html

--
Greg A. Woods

+1 416 218-0098; <g.a.woods@ieee.org>; <woods@robohack.ca>
Planix, Inc. <woods@planix.com>; VE3TCP; Secrets of the Weird <woods@weird.com>

On Tue, 2003-03-04 at 16:18, Greg A. Woods wrote:

> Indeed.
>
> Time to (again?) post the link to circuit diagram of what I believe is
> an equivalent device.
>
> The diagram was originally posted in sun-spots v5n45 by Malcolm Harper,
> back before the WWW and most copies are in compressed archives and thus
> not indexed by the likes of Google. A copy can now be found at this
> location:
>
> http://sunsite.kth.se/sun/sun-spots/v5n45.Z


Awesome read. The schematics talk about what happens when a cable is
unpluged from ttya. Is this still an issue or the same issue as the
break?


>
> The relevant post has been re-copied to at least these locations
> subsequently:
>
> http://sunsite.bilkent.edu.tr/pub/sun-info/sunspots/v9n20
> http://mail-index.netbsd.org/port-sparc/2002/05/16/0000.html
>
> --
> Greg A. Woods
>
> +1 416 218-0098; <g.a.woods@ieee.org>; <woods@robohack.ca>
> Planix, Inc. <woods@planix.com>; VE3TCP; Secrets of the Weird <woods@weird.com>
> _______________________________________________
> users mailing list
> users@conserver.com
> https://www.conserver.com/mailman/listinfo/users

On 4 Mar 2003, Chris Fowler wrote:

> On Tue, 2003-03-04 at 16:18, Greg A. Woods wrote:
>
> > Indeed.
> >
> > Time to (again?) post the link to circuit diagram of what I believe is
> > an equivalent device.
> >
> > The diagram was originally posted in sun-spots v5n45 by Malcolm Harper,
> > back before the WWW and most copies are in compressed archives and thus
> > not indexed by the likes of Google. A copy can now be found at this
> > location:
> >
> > http://sunsite.kth.se/sun/sun-spots/v5n45.Z
>
>
> Awesome read. The schematics talk about what happens when a cable is
> unpluged from ttya. Is this still an issue or the same issue as the
> break?
>
yup.
But, if the device is listed at
http://www.conserver.com/consoles/breakinfo.html as passing the test, then
it most likely will not result in a Sun at PROM. We have Cisco 3640s.
You can power cycle them until the cows come home and the Sun's don't
care. But, you can still send a break when you need to.