Terry Chan tpchan at
Mon Mar 1 14:58:34 GMT 2004

And it should be pointed out that only Couannette has had this exact problem
so far.  Several lunar devs have upgraded to openssh-3.8p1 on various
hardware configurations with NO problems.  Configurations tested included,
gcc-3.2.3, gcc-3.3.3, kernel-headers 2.4.x and kernel-headers 2.6.x.  So unless
someone else can duplicate this problem, it will be very difficult to track
it down and attempt to fix it.

Terry Chan
On Mon, Mar 01, 2004 at 08:10:03PM +0100, Couannette wrote:
> Hi,
> OpenSSH gets updated to version 3.8p1 in moonbase.
> Shortly after sshd begins to die unexpectedly. This is, IMHO a key
> problem. I believe that sshd/ssh have different algorithms to read/test
> server and user keys.
> I read the OpenSSH FAQ, and A/Q #2.6 and #2.6 made their way into my
> mind, and that's why I pretend there is a key bug/mismatch somewhere.
> I worked on with tchan and after cleaning my user keys, ssh would work
> again. But the daemon still died. It failed to read/test host keys in
> /etc/ssh.
> I tested a more aggressive cleasing: re-creation of my server keys. But
> no change. Sshd still died.
> I recreate now the file /etc/ssh/moduli to build new key hereafter. But
> it takes huge amount of proccessing time !!! When it finish I'll hold on...
> We must ensure proper operation of this, somewhat ugly, OpenSSH release.
> I'll continue to work on this topic as needed.
> Couannette

More information about the Lunar mailing list