poor performance from ssh 1:3.6.1p2-4 backported to woody

[Cristian Ionescu-Idbohrn <cristian.ionescu-idbohrn@axis.com>]

Hi,

I needed to upgrade ssh on our cvs-server because of this:

  * Backport upstream patch to pass monitor signals through to child
    (closes: #164797).

We had lots of gost processes left over because signals were not
propagated correctly. The problem is still not completelly cured. I
noticed today that a:

  # cvs up

which failed with a "disk full" error left 4 processes (2 ssh ad 2
cvs) on the cvs-server.

After the uppgrade I also noticed poor performance when starting a ssh
session. A 3.6.1p2 or 3.4p1 client to a 3.4p1 server is almost
instantaneous. OTOH, a 3.6.1p2 or 3.4p1 client to a 3.6.1p2 server
takes 2-3 seconds or more to login. Thus the slowdown when running
several sequential logins _is_ painful.

Anyone else noticed this behaviour? Is there anything one could do to
speed it up?


Cheers,
Cristian