Re: long daemon outages on upgrade
In article <[🔎] 87n0xcl7bp.fsf@rover.gag.com> bdale@gag.com writes:
>The 'bind' packages which deliver version 8.X bits have been hand-holding the
>daemon restart in the postinst to minimize downtime since version 8.2.2p5-2
>uploaded in November of 1999. I introduced a bug into this by accident a
>while back that LaMont fixed for our 8.3.1-1 upload, so if it's not doing the
>right thing now that's a new bug.
And the hand-holding code introduces a security problem by not
restarting named with the proper options. (bug #135991) Frankly, I'd
rather have named down for a few minutes on upgrade rather than have
it start without the -u option. Using "/etc/init.d/bind start" for the
restart would fix this problem.
Reply to: