Re: Proper way to run bind9

From: Doug Barton (DougB_at_FreeBSD.org)
Date: 09/25/04

  • Next message: Kris Kennaway: "Re: 5.3 IPSEC broken"
    Date: Sat, 25 Sep 2004 00:44:35 +0100 (BST)
    To: Sean McNeil <sean@mcneil.com>
    
    
    

    On Fri, 24 Sep 2004, Sean McNeil wrote:

    > On Fri, 2004-09-24 at 14:27, Doug Barton wrote:
    >> On Fri, 24 Sep 2004, Dag-Erling Smørgrav wrote:
    >>
    >>> Grover Lines <grover@ceribus.net> writes:
    >>>> named_pidfile="/var/run/named/pid" # Must set this in named.conf as well
    >>> ^^^^^^^^^^^^^^^^^^
    >>> should be /var/run/named.pid, fixed in CVS.
    >>>
    >>> DES
    >>>
    >>
    >> It's actually not named.pid in our structure. As explained in the note
    >> behind the variable, we set the pid-file variable in named.conf so that
    >> named running wit h -u bind (but not chrooted) will still be able to
    >> drop a pid file in /var/run/named, which is chowned to user bind.
    >
    > This is currently not correct in some files (i.e.
    > /etc/defaults/rc.conf).

    DES made an honest mistake in rc.conf. I was waiting for him to back it
    out himself, but apparently he's off having a life or something. :) I
    went ahead and fixed it a bit ago.

    You mentioned "some files," do you know of anywhere else that it is not
    correct?

    > Also, the /etc/rc.d/named script will do an
    >
    > ln -fs "${named_chrootdir}${pidfile}" ${pidfile}
    >
    > if named_symlink_enable is set (which is by default). Please protect
    > this with
    >
    > if [ -n "$named_chrootdir" ]; then

    On my system this fails harmlessly, but you're right, it shouldn't be
    run if there is no chroot. I'll commit a fix for this in a second.

    > for those who do not have a chrootdir. Otherwise we end up with a
    > recursive link.

    Actually the final result of this situation depends on the configuration
    options. But you're right, this bullet should be removed from the
    foot-shooting gun.

    >> To answer Grover's question, it really depends on what you want to use
    >> it for. The system named.conf will run fine for bind 9 as a resolver,
    >> now that the /etc/rc.d/named script has been updated to create an
    >> rndc.key file if one doesn't exist.
    >
    > This is broken too. If named_chrootdir isn't set, then confgen_chroot
    > doesn't get set and it messes up the invokation of rndc-confgen. I
    > think taking the "" off of the ${confgen_chroot} will solve this but I'm
    > not sure.

    Only broken for the non-chroot case. :) You're right about the fix
    though, I committed the wrong version when I was testing it last night.
    Thanks for the bug report.

    >> If all you want to do is start up named as a resolver,
    >> named_enable="yes" is all you need. You don't need to specify the conf
    >> file to run the system's version of bind, that path is defined in.
    >>
    >> I'm currently working on a setup so that named can be started chrooted
    >> by default. Not sure if that will get in before 5.3-RELEASE or not, but
    >> I'm hoping it will.
    >
    > It would be nice to have it all working while you make these changes.

    That is the goal, yes.

    Thanks again,

    Doug

    -- 
         This .signature sanitized for your protection
    
    

    _______________________________________________
    freebsd-current@freebsd.org mailing list
    http://lists.freebsd.org/mailman/listinfo/freebsd-current
    To unsubscribe, send any mail to "freebsd-current-unsubscribe@freebsd.org"


  • Next message: Kris Kennaway: "Re: 5.3 IPSEC broken"

    Relevant Pages

    • cvs-src summaries misposted
      ... about the fact that I wouldn't be able to post the cvs-src summaries. ... Søren Schmidt added support to the ATA code for detaching PCI ATA ... As a follow-up to last week's commit of `ICC support for kernel building`, ... imported OpenSSL 0.9.7d, which includes the fix, to -CURRENT. ...
      (freebsd-current)
    • Re: [PATCH 25/39] merge common parts of uaccess.
      ... commit is rather large. ... I'm sure the fix is simple, but this bug shows a structural problem ... bisection result - instead of this too-broad bisection result. ... final step can be mechanic in that it unifies exactly the same content ...
      (Linux-Kernel)
    • Re: [PATCH] blk: missing add of padded bytes to io completion byte count
      ... I think that the block layer had better take care about it (fix ... But I plan to send a patch to revert it ... No this commit is a serious bug, and the only fix is like you suggested ...
      (Linux-Kernel)
    • [GIT PATCH] ACPI patches for 2.6.25-rc5
      ... ACPI: button: make real parent for input devices in device tree ... acer-wmi: Rename mail LED correctly & remove hardcoded colour ... Fix a duplicate log level ... commit 29ea5171cbd08a7ef37e543cbf02447033f0e289 ...
      (Linux-Kernel)
    • Re: [GIT PATCH] another tranche of SCSI updates for 2.6.26
      ... commit 064922a805ec7aadfafdd27aa6b4908d737c3c1d ... This patch adds more const keywords where appropriate. ... fix SLUB WARN_ON ... KVM: SVM: remove now obsolete FIXME comment ...
      (Linux-Kernel)