Stephen Frost <sfrost@snowman.net> [2002-09-16 16:10:58 -0400]: > * Bob Proulx (bob@proulx.com) wrote: > > Agreed. I consider bind locally a standard installation. Over a > > There are alot of reasons to not one run too. Agreed. There is no such thing as "one size fits all". However there are sometimes only "one size". Trying to avoid that case. > There is no overriding reason or need for named to have an entry in > base-passwd. It's simple enough for the package to add the user. > Even ssh does this, and for much the same reason. Note that I was only stating that a local named is very common and was refuting the statements that no one runs one. I expressed no opinion on whether 'named' should be in base-passwd or not. As long as I am here, previous mention of keeping uids consistent over NFS, I believe anyone that is running NFS and wants uids to be consistent across hosts will likely also be running NIS/YP as well. Not requied but probably very likely. In that case, I just checked, if an NIS entry exists for the user you are trying to add then adduser will complain and exit non-zero. The 'getent' program will report the user from NIS. There is undoubtedly a way to install, detect 'named' in NIS and react accordingly. Bob
Attachment:
pgpKUw7JLCP7c.pgp
Description: PGP signature