Date: Sat, 13 Oct 2001 18:51:49 +0200 (CEST) From: "Hartmann, O." <ohartman@klima.physik.uni-mainz.de> To: <freebsd-stable@freebsd.org> Cc: <freebsd-bug@freebsd.org> Subject: SSH broken in FBSD 4.4-STABLE!! Message-ID: <20011013181212.R22627-100000@klima.physik.uni-mainz.de>
next in thread | raw e-mail | index | archive | help
Dear Sirs. Maybe you remember my several postings in the past, I found out some strange things happening to sshd when migrating from FBSD 4.3 to 4.4. Now I have results, and results I can reproduce! But first I will explain the history. In the early days, before FreeBSD went towards 4.4-RC2, sshd worked perfectly on our systems. SSH is on all systems started in the normal, recommended way by the /etc/rc.conf.local config file. The first symptomes were that sshd stated after reboot or not, while moving tighter towards 4.4-RELEASE and now to 4.4-STABLE, sshd never started after rebooting, it always reports this error in thelog: Oct 12 22:23:25 <auth.crit> klima sshd[289]: fatal: rsa_generate_private_key: key generation failed. The only way to start sshd was to telnet to the appropriate host and start the daemon manually. And this manual start does not produce any problem! The I started sshd via rc.local - not problem. It seems, that it depends on __when__ sshd starts in the rc-triggered startup process. I administer at this moment a core of four servers around here: three SMP machines, one UP machine. The failing start of sshd is watched on two SMP machines - both systems has Pentium PIII Coppermines! The other SMP system has two older PII CPUs. The UP machine is based on an Athlon TBird 850. Today I tryed to find out what the difference could be between the SMP machined that fail, the one MP machine that fails not and the UP system. Well, the ENABLE_SSE switch in the kernel causes the problems in conjunction with a real SSE capable Intel PIII CPU! I have no chance to test this on an PIII/SSE UP machine, but on both SMP machines using PIII and enabled SSE switch in kernel sshd is not starting with the normal rc script, but it starts the normal way if ENABLE_SSE is commented out! On the PII box, this switch has no effect. This is the same on the UP machine. ENABLE_SSE gives us several performance benefits (we use PGI and Lahey Linux Fortran and they seem to run about several percent faster on boxes with enabled SSE, doing scientific calculations) and I do not want to miss it. It is curious that the starting order of sshd is depend whether sshd starts or not. Maybe someone capable to do that should start sshd later in rc. Please, someone out there should test and confirm my experiences and if it is not SSE specific (but I presume it's a fact), I have to look for other problems within my configuartions (but this is logicaly very unlikely ...). Could anyone test this with an PIII UP System? Thank you for your help. -- MfG O. Hartmann ohartman@klima.physik.uni-mainz.de ---------------------------------------------------------------- IT-Administration des Institutes fuer Physik der Atmosphaere (IPA) ---------------------------------------------------------------- Johannes Gutenberg Universitaet Mainz Becherweg 21 55099 Mainz Tel: +496131/3924662 (Maschinenraum) Tel: +496131/3924144 FAX: +496131/3923532 To Unsubscribe: send mail to majordomo@FreeBSD.org with "unsubscribe freebsd-stable" in the body of the message
Want to link to this message? Use this URL: <https://mail-archive.FreeBSD.org/cgi/mid.cgi?20011013181212.R22627-100000>