Re: Blocked high ports
Agreed on his problem, but I HAVE seen a bit of instability during several
potato upgrades. Generally the system will run properly, but I've had times
where things break majorly until I reboot.
Dave Bristel
On Tue, 25 Jan 2000, Craig Sanders wrote:
> Date: Tue, 25 Jan 2000 10:59:52 +1100
> From: Craig Sanders <cas@taz.net.au>
> To: David Bristel <targon@targonia.com>
> Cc: Bradley M Alexander <storm@tux.org>, debian-devel@lists.debian.org
> Subject: Re: Blocked high ports
>
> On Mon, Jan 24, 2000 at 11:35:19AM -0800, David Bristel wrote:
> > Have you checked to make sure that inetd is running? After an
> > upgrade, be sure to reboot to make sure that library differences have
> > taken effect, and to make sure the system is in sync. I have found
> > that in too many cases, things LOOK fine after an upgrade, but until
> > that final reboot, it won't be fully cut-over to the new version.
> >
> > Dave Bristel
>
> i've been upgrading debian on dozens of systems for nearly 5 years and
> have NEVER noticed what you describe. in fact, smooth upgrades where
> everything *just continues working* after being upgraded with no hassle
> and no fuss is one of debian's major benefits.
>
> i upgrade my systems regularly (every week or two on average) and
> generally only reboot for hardware changes (once every few months at
> most) or because of kernal bugs.
>
> what you suggest is not necessary - it's a windows-type approach to
> computing: "reboot and it might fix itself" (it's also a slackware and
> freebsd type approach because they use ugly monolithic rc scripts,
> rather than the clean sysvinit scripts).
>
> debian packages provide and use scripts in /etc/init.d to stop
> and re-start the daemons as required during an upgrade. rebooting
> is not required - in fact, it's so far from being "required" that
> rebooting should be one of the last things you consider when trying to
> diagnose/solve a fault. rebooting doesn't solve anything, at best it's
> cargo-cult systems administration...perform the magic reboot and maybe
> the problem will go away for a while. the trouble with that is that
> you still don't know what *caused* the problem and you have no way of
> knowing that it isn't going to happen again.
>
> aside from kernel upgrades, the only time that a reboot has ever been
> necessary when upgrading debian was during the libc5 to libc6 migration,
> which was (as everyone who went through it knows) a major change....and
> even that was optional rather than required if you didn't care about the
> wtmp file being corrupted because the file format changed. it was highly
> recommended to reboot, but not absolutely required.
>
>
> it is far more likely that Bradley's problem is due to incorrectly set
> firewall or masquerading rules or similar. or a misconfigured network
> interface, or some other misconfiguration.
>
> craig
>
> --
> craig sanders
>
Reply to: