Code review comment for lp:~smoser/ubuntu/oneiric/isc-dhcp/lp857524

Revision history for this message
Scott Moser (smoser) wrote :

I suppose, yes. There is the potential for something to race on that file.
The old code path (and with a move) would still potentially have a race conditoin when it tries to read /etc/resolv.conf and get the old name servers, and then uses those for updating. Something could edit in between.

All that said, this should be *really* fast.
neither the DHCPv6 or DCHPv4 paths do any IO during the writing of that file.
Its all just shell builtins and variable operations.

You're correct, though, there is the potential for race. Should this be NAK'd because of that?

« Back to merge proposal