Date: Fri, 28 Jan 2005 18:44:46 -0800 From: Brooks Davis <brooks@one-eyed-alien.net> To: Mike Makonnen <mtm@identd.net> Cc: rc@freebsd.org Subject: Re: review request: early rc.d scripts Message-ID: <20050129024446.GA21989@odin.ac.hmc.edu> In-Reply-To: <20041014100429.GA1626@rogue.acs.lan> References: <20041008223214.GA4501@odin.ac.hmc.edu> <20041014100429.GA1626@rogue.acs.lan>
next in thread | previous in thread | raw e-mail | index | archive | help
--0OAP2g/MAC+5xKAE Content-Type: text/plain; charset=us-ascii Content-Disposition: inline Content-Transfer-Encoding: quoted-printable On Thu, Oct 14, 2004 at 01:04:30PM +0300, Mike Makonnen wrote: > Sorry for the late reply. I meant to reply to this sooner but > forgot about it. Some how I missed this reply back in October and then got distracted from this work. > I generally don't like multiple runs of rcorder(8) since this > can potentially make the system fragile. This is because an early script > gets run twice, and if the script doesn't do sanity checking the second > time round it could break things (rc.d/var comes to mind). I realize that > the keyword is currently only in two rc.d scripts, but having the > functionality available invites use in other scripts. The scripts at NOT run twice because they are skipped by "-s early" being added to the skip variable. There's a slight cost in rescanning /etc/rc.d twice but in the common case everything will be in the cache the second time through. > Secondly, this seems to me to be of limited use to the majority of > FreeBSD users. I can't think of a single instance in the past where > I wished I had this functionality. However, you certainly want it, so > there will probably be others that could use it too :-) Given teh minimal cost, I'm not sure I buy this. > Given my reservations I would really prefer you keep > it as a local hack or find a different mechanism. However, if you really > want it in the base system it should be behind an rc.conf knob that > defaults to off so that those people who don't use it won't be > affected by it. I think something like this should be in the base system because the current situation badly violates POLA. You can put rc.d scripts under /conf and you'll see them when you log in, but they don't run. The real problem is that initdiskless probably should not have been made in to an rc.d script in the first place. It's setting up /etc so the rcorder run is too early. I can't put this under an rc.conf variable know because there may well not be an /etc/rc.conf in existance at this point. An alternative would be to move initdiskless back to /etc and run it explicitly before rcorder. That might be the best approach. -- Brooks --=20 Any statement of the form "X is the one, true Y" is FALSE. PGP fingerprint 655D 519C 26A7 82E7 2529 9BF0 5D8E 8BE9 F238 1AD4 --0OAP2g/MAC+5xKAE Content-Type: application/pgp-signature Content-Disposition: inline -----BEGIN PGP SIGNATURE----- Version: GnuPG v1.2.1 (GNU/Linux) iD8DBQFB+vibXY6L6fI4GtQRAuWGAJ4xFg8ZujIDm+BgSidQAPxGnrsYvACguTha +J/swTxGle9R6hBkgANOpb8= =gPEz -----END PGP SIGNATURE----- --0OAP2g/MAC+5xKAE--
Want to link to this message? Use this URL: <https://mail-archive.FreeBSD.org/cgi/mid.cgi?20050129024446.GA21989>