Bug#205039: libc6: don't just ask if I want to restart all services on installation

[GOTO Masanori <gotom@debian.or.jp>]

severity 205039 wishlist
thanks

At Tue, 12 Aug 2003 08:27:13 +0800,
Dan Jacobson wrote:
> Upon installation, I see:
> 
>     Setting up libc6 (2.3.2-2) ...
>     Checking for services that may need to be restarted...done.
> 
>     Name Service Switch has changed in the C Library: post-installation question.
>     Running programs may not be able to do NSS lookups until they are
> 
> 
> here say what NSS is, so users don't know
> 
> 
>     restarted (for services such as ssh, this can affect your ability to
>     login).
>     Note: restarting sshd/telnetd should not affect any existing connections.
> 
>     The services detected are: inetd spamassassin cron apache ssh 
> 
>     If other services begin to fail mysteriously after this upgrade,
>     it may be necessary to restart them too.  We strongly recommend
>     you to reboot your machine to avoid the NSS related trouble.
> 
> OK, I better reboot, even though I don't know what NSS is.
> 
>     Do you wish to Restart Services? [Y/n] 
> 
> Why don't you give a third choice: "ask which ones",
> don't just assume that if one is installed, we run it all day
> 
> See e.g. http://groups.google.com/groups?threadm=jq3M.1pf.5%40gated-at.bofh.it
> ("better make a standard for /etc/*/*_not_to_be_run")

I don't read this thread (if you can, could you summarize this
thread?), but I think it's /etc/init.d or /etc/rc?.d/ work.  NSS
breakage needs to restart all services.  But we don't know it's
working or not.  Only we can do is restarting /etc/init.d/ scripts.

I don't think it's anytime good idea to add "ask which ones", because
most users don't know what NSS is, and it must restart for such users
without selection.  Controlling daemon activities should do modifying
/etc/rc scripts.

However, it's happy if we can add postgresql entry using with your
proposed selection mechanism.  postgresql is transaction-oriented, and
if we stop immediately using /etc/init.d, then transactions start
rolling back.  So we're dropping postgresql from restarting scripts.
However, if user can select to restart or not, it's valuable.
Could you contribute such code? :)

Regards,
-- gotom