Mailing List Archive

nope...that's one design decision of conserver that i really like...only
one person typing at any one point. you can use the ^Ecf option to
steal the console away at any time (so that your typing hits the
console), but not at the same time. is there some reason you want
multiple writers at once? or is it just a habit that you might be able
to change? if you really want multiple writers, you can always patch
the code...there's probably only one or two places something would need
to be commented out/adjusted.

Bryan

On Tue, Mar 07, 2006 at 11:37:19AM -0800, Brian Matheson wrote:
>
> Hi all,
>
> Is there a way to make conserver allow multiple read/write sessions to a
> single console? I'm looking to use conserver to duplicate the
> functionality of Cyclades ACS systems and this is sort of the last
> feature I need to make it a drop-in replacement.
>
> Thanks much,
> Brian
>
> BTW, the Raritan ads were hysterical.
> _______________________________________________
> users mailing list
> users@conserver.com
> https://www.conserver.com/mailman/listinfo/users
_______________________________________________
users mailing list
users@conserver.com
https://www.conserver.com/mailman/listinfo/users

i should also say...i can't guarantee all the code will work with
multiple writers. there are some features (like ^Ec|) that would
probably break badly...and possibly others. there was an inherent
assumption that only one writer exists at a time and the code does it's
best to both enforce and take advantage of that fact. it would require
a full review to really see if those assumptions are there, but my
instinct says things probably wouldn't work - and in bizarre ways.

Bryan

On Tue, Mar 07, 2006 at 11:47:32AM -0800, Bryan Stansell wrote:
> nope...that's one design decision of conserver that i really like...only
> one person typing at any one point. you can use the ^Ecf option to
> steal the console away at any time (so that your typing hits the
> console), but not at the same time. is there some reason you want
> multiple writers at once? or is it just a habit that you might be able
> to change? if you really want multiple writers, you can always patch
> the code...there's probably only one or two places something would need
> to be commented out/adjusted.
>
> Bryan
_______________________________________________
users mailing list
users@conserver.com
https://www.conserver.com/mailman/listinfo/users

Harris, David (SBS US) <david.k.harris@siemens.com> wrote:

> Hmmm...why the multiple read/write sessions?

In large part because we have systems that currently work that way. I
actually like the idea of having only one writer at a time, but I think
it'll be a hassle to get my users into the habit of doing that.

> Are you suggesting that anyone could type at any time? Or do I
> misunderstand? Could you describe how you think this would work, in
> your ideal world?

Pretty much anyone can type at any time. In practice, folks leave
sessions open for a long time, and might have multiple terminals
attached to them from different machines. Maybe worth noting that these
are non-production systems, and there should only be one primary user at
a time. We might get burned, but I haven't had any complaints about it
so far.

Bryan Stansell <bryan@conserver.com> wrote:

> if you really want multiple writers, you can always patch
> the code...there's probably only one or two places something would need
> to be commented out/adjusted.

Cool, it's good to know that you don't see any major barriers to doing
it. I'm no c coder, but I did peek around in the sources a bit trying
to find any gotchas. I'll see if I can find and remove the checks for
read/write permissions and get it to build :^)

Thanks much for the info,
Brian


_______________________________________________
users mailing list
users@conserver.com
https://www.conserver.com/mailman/listinfo/users

>>>>> "Brian" == Brian Matheson <bmath@xsigo.com> writes:

Brian> Harris, David (SBS US) <david.k.harris@siemens.com> wrote:

>> Hmmm...why the multiple read/write sessions?

Brian> In large part because we have systems that currently work that
Brian> way. I actually like the idea of having only one writer at a
Brian> time, but I think it'll be a hassle to get my users into the
Brian> habit of doing that.

Just get your users used to doing the ^Ef key to force writeability
and they'll be just fine with it. And they'll appreciate the fact
that only one terminal at a time can write.

The big thing I want is a bigger scroll back buffer. The default 20
and 60 lines aren't enough at all. I'd like 10-60 screens of lines of
scollback if possible.

John
_______________________________________________
users mailing list
users@conserver.com
https://www.conserver.com/mailman/listinfo/users

On Tue, Mar 07, 2006 at 05:03:25PM -0500, John Stoffel wrote:
> Just get your users used to doing the ^Ef key to force writeability
> and they'll be just fine with it. And they'll appreciate the fact
> that only one terminal at a time can write.
>
> The big thing I want is a bigger scroll back buffer. The default 20
> and 60 lines aren't enough at all. I'd like 10-60 screens of lines of
> scollback if possible.

now, *that's* easy to hack into the code (if you want to "go there").
look for Replay() in conserver/group.c...just change the 20 and 60 to
whatever you prefer. i still need to work on getting that more
"dynamic" and/or customizable...

Bryan
_______________________________________________
users mailing list
users@conserver.com
https://www.conserver.com/mailman/listinfo/users

The *BIG* advantage to using ^E-c-f is that you end up with
a stamp in the log about who was in control, and that can be
useful sometimes, in helping to find an admin who needs some
'additional clue(s)'. :-)

-Z-

-----Original Message-----
From: John Stoffel [mailto:john.stoffel@taec.toshiba.com]
Sent: Tuesday, March 07, 2006 2:03 PM
To: Brian Matheson

>>>>> "Brian" == Brian Matheson <bmath@xsigo.com> writes:
Brian> In large part because we have systems that currently work that
Brian> way. I actually like the idea of having only one writer at a
Brian> time, but I think it'll be a hassle to get my users into the
Brian> habit of doing that.

Just get your users used to doing the ^Ef key to force writeability and
they'll be just fine with it. And they'll appreciate the fact that only
one terminal at a time can write.

The big thing I want is a bigger scroll back buffer. The default 20 and
60 lines aren't enough at all. I'd like 10-60 screens of lines of
scollback if possible.

John

_______________________________________________
users mailing list
users@conserver.com
https://www.conserver.com/mailman/listinfo/users

On Tue, Mar 07, 2006 at 11:37:19AM -0800, Brian Matheson wrote:
> Is there a way to make conserver allow multiple read/write sessions to a
> single console? I'm looking to use conserver to duplicate the
> functionality of Cyclades ACS systems and this is sort of the last
> feature I need to make it a drop-in replacement.

You could use the 'screen' program in front of console sessions.

www.gnu.org/software/screen

Matt

_______________________________________________
users mailing list
users@conserver.com
https://www.conserver.com/mailman/listinfo/users

Yes...but screen is only watching when I'm connected...

With Conserver, the app is 'always watching', always logging.
When I connect to Conserver using the client, a note goes into
the log for the machine that I'm conencting to...and when I
disconnect from Conserver.

When I'm connected, and you also connect, you are in read-only
mode, because I've got the write access...if you force-connect;

+ the log shows that you took the write access
+ my client shows that I was "bumped by matt_hannigan"
+ I get to see what you did in my scrollback

When you disconnect from Conserver, write control reverts
back to me, and another timestamp goes in the log. Even if
you simply go into 'spy mode', the write control reverts,
and the log entry is made.

The write access is heirarchical...you steal it from me,
Bob steals it from you, Bryan takes it from Bob...the stack
now shows zonker -> matt -> bob -> bryan... If Bob disconencts,
and then Bryan disconnect, it reverts directly to you, as you
are the next highest in the stack.

The model here is;

I get on a console, hack a bit, and then I get pulled into a
meeting, or go to lunch...

You need to get on the console...you *CAN*, in spy mode. BUT,
you also know who has write access ("no, zonker is connected").
You can call me, send me email, walk to my cube...

^E-c-r replays the recent log (was I in the middle of a config?
You don't just want to take control and hit [return] a few times!

^E-c-w shows who else is on the port, and how long we've been idle.
(Have I gone home, or was I working on it a few minutes ago?)

Can't reach me, and you *really* NEED write access? ^E-c-f, takes
control, and you make your changes, and then ^E-c-s (spy mode) or
^E-c-. (disconnect), and I'm back where I was when you are done.
(BUT, the master log shows that this has occurred.)

The logging is important. Sometimes I'm monitoring a port from home,
via a VPN, and the link dies (or times out). I don't know what is
happening, and TCP still thinks my session is up. You are not impacted,
and can still do what you need to do. :-)

-Z-

-----Original Message-----
From: users-bounces@conserver.com [mailto:users-bounces@conserver.com]
On Behalf Of Matthew Hannigan
Sent: Tuesday, March 07, 2006 3:34 PM
To: Brian Matheson
Cc: users@conserver.com
Subject: Re: Multiple rw sessions?

On Tue, Mar 07, 2006 at 11:37:19AM -0800, Brian Matheson wrote:
> Is there a way to make conserver allow multiple read/write sessions to

> a single console? I'm looking to use conserver to duplicate the
> functionality of Cyclades ACS systems and this is sort of the last
> feature I need to make it a drop-in replacement.

You could use the 'screen' program in front of console sessions.

www.gnu.org/software/screen

Matt

_______________________________________________
users mailing list
users@conserver.com
https://www.conserver.com/mailman/listinfo/users

_______________________________________________
users mailing list
users@conserver.com
https://www.conserver.com/mailman/listinfo/users