[25873] in Source-Commits
Re: /svn/athena r25272 - trunk/debathena/scripts/installer/pxe/stage1/debathena
daemon@ATHENA.MIT.EDU (Jonathan Reed)
Tue Jul 26 09:29:49 2011
Mime-Version: 1.0 (Apple Message framework v1084)
Content-Type: text/plain; charset=us-ascii
From: Jonathan Reed <jdreed@MIT.EDU>
In-Reply-To: <alpine.DEB.2.00.1107260904210.17504@tyger.mit.edu>
Date: Tue, 26 Jul 2011 09:29:41 -0400
Cc: source-commits@mit.edu
Message-Id: <71E78E51-F2FB-4B5E-976C-6D268BF8E7F1@mit.edu>
To: Geoffrey Thomas <geofft@mit.edu>
Content-Transfer-Encoding: 8bit
On Jul 26, 2011, at 9:20 AM, Geoffrey Thomas wrote:
> Bah.
>
> 1. It might make a lot more sense to use the installer's existing syslogging support instead of writing our own log. (As a side benefit, syslogs are timestamped.) You can use what appears to be a reasonably standard logger(1) to send messages. They end up in /var/log/syslog.
Ah, I didn't realize we had "logger" at this point.
>
> 2. I'm skeptical that your netcat MTA works. The client not waiting for server responses is one of the biggest giveaways that a message is spam before it's even read; I suspect this will be highly unreliable. The installer has a built-in httpd for getting debug logs off -- you can just run "httpd", which will spawn a daemon serving /var/log over port 80, or you can do something involving `udpkg --configure --force-configure save-logs`, which also symlinks a few useful things into /var/log. (The latter is what the installer main menu's "Save debug logs" option runs.
It worked repeatedly in my testing, but httpd is possibly a better option. However, I'd like something that doesn't require the user to keep the machine sitting there forever until someone from release-team can look at it.
>
> 3. If networking doesn't work, how is sending mail supposed to work?
If networking doesn't work, how is httpd supposed to work?
Even with an 18.2 address, you can talk to 18/8.
I'll have a second draft shortly.
-Jon
