[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]
Re: [sysvinit-devel] New beta release
From: |
Jesse Smith |
Subject: |
Re: [sysvinit-devel] New beta release |
Date: |
Mon, 20 May 2019 13:38:50 -0300 |
User-agent: |
Mozilla/5.0 (X11; Linux x86_64; rv:60.0) Gecko/20100101 Thunderbird/60.6.1 |
>> In particular, startpar now loads Makefile-style dependency information
>> from /var/lib/insserv/ instead of /etc/init.d/. It is possible to
>> override this and use the old location. Likewise, insserv now defaults
>> to writing Makefile-style dependency information to /var/lib/insserv/.
>> Again, the location can be changed back to the old /etc/init.d directory
>> if need be. In both programs the command line flag to use the classic
>> location is "-l" (lower-case L).
> This sound like a good location for the files, but I suspect it is not.
>
> The main reason these files were loaded from a directory under / and not
> /var/, is that /var/ is not guaranteed to be mounted when the files need
> to be read. Why do you believe this has changed?
First, I don't believe it has necessarily changed, which is why startpar
checks both the new and legacy locations for the files it needs. If /var
is not mounted, it won't cause an error, we'll just fall back to using
/etc. Likewise, people with detached /var filesystems can still run
insserv with the legacy location. If /var is not mounted in time there
should be no issue.
Second, this change was initiated by, and debated, downstream. They feel
it's safe to assume /var is available when startpar and/or insserv run.
However, they do not believe it is safe to assume /etc is mounted
read-write (which I agree with). Hopefully this beta release will prove
the issue one way or the other.
Jesse