[Date Prev] [Date Index] [Date Next] [Thread Prev] [Thread Index] [Thread Next]

Re: Further break-sequence weirdness

Bryan Stansell bryan@conserver.com
Tue, 21 May 2002 12:42:07 -0700 (PDT)

hi aaron,

On Tue, May 21, 2002 at 10:55:32AM -0700, Aaron Burt wrote:
> When I run the console client on Build, I can use ^Ecl1 to send a break to
> boxes on Term00, but not Term01.

your first post about this made me think there might have been a
configuration issue on Term01...but considering the following:

> When I run the console client on Octopus, connecting to Build, I can use
> ^Ecl1 to send a break to boxes on Term01, but not Term00.

i kinda doubt it.

> Weird, huh?

more like impossible, actually (my immediate reaction), but who knows.
so, the assumption i'm reading in this is that Build is the only host
actually running the conserver binary, right?  it's making the tcp
connections to both Term00 and Term01.  the only role in this that
Octopus plays is running the client (and as you said, connecting to
Build).  if this is true, then maybe you just "got lucky" and saw a
pattern where there really isn't one?  i say that because all the
logic, all the break sending, all the "everything important" is
happening inside the conserver binary - the client is just triggering
events inside it.  so, regardless of where the client lives, if the
break sequence is triggered in a particular conserver binary, it
*should* behave the same every time (hence my immediate "impossible"

but something is obviously wrong, and it's weird enough that i'll need
a detailed explaination of the setup (or a confirmation that my
assumptions above are correct - that's probably enough detail - and
maybe a copy of the conserver.cf file would help too) and serious
amount of debug output.  if you run both conserver and the console
client with -DD (yep, two D's) on all the hosts and redo your tests,
maybe i'll be able to see something.  it *will* be a lot of output.

and you may want to connect multiple times from Octopus or Build and
see if that changes the behavior - or even open another window and bump
yourself off from the same client host and see.  basically, i'm
wondering if another pattern can be found.

you know where to find me if you're able to put together all that
info.  :-)