Aaron, Aaron Morris writes: > rsync is just doing what your are telling it to do. Update if the file > is changed or does not exist on the remote side and delete if it no > longer exists on the local side.
Yes, I can RTFM, and, suprisingly!, I even _did_ RTFM :)). > The directory may have a newer > timestamp, but you are doing a recursive put so it has to check all the > files and dirs under each directory. At least that is how I see it. > > solution: --exclude "/tmp/" or --exclude "tmp/" Well, I think you missed the point. The "tmp" subdir was just an example. The same can happen in any subdir down the tree. And, if you re-read my post carefully (or even the subject), I said "--update" (mis)behavior __on_directories__ (and _not_ regular files). Truth to be told, after a bit of thinking I _might_ accept the fact that "--update" ignores the timestamps on (sub)directories. Because if it didn't, if any file was "touch"-ed remotely after the last rsync than all subdirs, up to and including to the "top" dir, would have a newer timestamp remotely and thus no file putting would take place. However, I still think that "--update" (or, ideally, another similar option or combination of options) __should__ take into account (sub)directories timestamps and consider for processing only those (sub)directories that have remotely an older timestamp than locally. Or, to put it in your words "...doing a recursive put should be restricted __only__ to the (sub)directories that remotely have a timestamp older than locally". If that means that _no_ update (i.e. put) will take place because __any__ file deep down the tree was touched remotely since the last put...well, than that's it. Again, thanks in advance to anyone can suggest a way to have rsync __also__ in this way. Best regards, Florian > Florian-Daniel Otel wrote: > > Dear all, > > > > I have the following problem: I use the following command to "push" > > files from "local" to a "remote" machine: > > > > [...] > > /usr/bin/rsync -avuz -e "ssh -1" --exclude ".Xauthority" --delete >/user/home/directory/ [EMAIL PROTECTED]:/user/home/directory/ ; > > > > (The ssh is using RSA authentication btw. machines, but that's beside the point). > > > > The problem: > > > > > > On both the local and remote machines there is a subdirectory > > "/user/home/directory/tmp". If on the remote machine that subdirectory > > has a newer time stamp -- e.g. due to a file that was created > > recently i.e. after the last time the local machine did a "pull" -- the > > "--delete" option in the command above makes it such that the said > > newer file is deleted on remote and the timestamp is reseted back to > > the old time stamp i.e. the one for the "tmp/" subdir on the local machine. > > > > > > Putting it shortly, the "--delete" option has precendence over "-u" > > option w.r.t. directories. > > > > Is it so that "-u" works only on files and not on directories ? > > IMHO the appropriate behaviour in a situation like the one described > > above directories on remote that have newer time stamp should be ignored > > and excluded from further processing (i.e. including updating and/or > > deleting any files that do not exist on the local machine). > > > > The question: > > > > Am I right and that is the way things should behave and I am doing > > smth wrong ? If not, and this is the way things are supposed to work, > > can anyone suggest a work-around that would simulate the desired behavior ? > > > > > > > > Many thanks in advance, > > > > Florian > > > > P.S. Please Cc: me on the replies, I'm not on the list. Thanks. > > > > P.P.S: If it matters: In the above setup the "local" machine is a > > Linux running "rsync version 2.5.4 protocol version 26" whereas > > "remote" is a Solaris 2.8 running "rsync version 2.5.5 protocol version 26". > > > > -- > Aaron W Morris > decep > PGP Key ID: 259978D1 > > -- To unsubscribe or change options: http://lists.samba.org/mailman/listinfo/rsync Before posting, read: http://www.tuxedo.org/~esr/faqs/smart-questions.html
