If rsync tries to sync a non existen directory it fails instead of ignoring it and continue. This breaks some scripts. For details see upstream upstream bugreport. Can you apply the suggested patch?
(In reply to M K from comment #0) > If rsync tries to sync a non existen directory it fails instead of ignoring > it and continue. This breaks some scripts. > > For details see upstream upstream bugreport. > > Can you apply the suggested patch? The author of that patch says: > Following patch fixes it for me, albeit it ignores this kind of error now > always: (See https://bugzilla.samba.org/show_bug.cgi?id=12569#c0 ) That sounds as if it also ignores such errors when --delete-missing-args and --ignore-errors are _not_ used. Assigning to all packagers collectively, since there is no registered maintainer for this package.
Keywords: (none) => UPSTREAMAssignee: bugsquad => pkg-bugsCC: (none) => marja11
We certainly shouldn't take such a patch if upstream hasn't either. Also, failing on missing directories is generally a good thing with rsync, as it helps prevent mistakes (such as doing an rsync involving a mounted filesystem that happens to not be currently mounted).
maybe you are right, but this is a changed behaviour which worked in mga5 and makes scripts fail, since there is no "force" available which lets you ignore it. For me a server creates files, they are periodically synced, but it can happen the file (dir) specified disappears before rsync can copy it. So the script fails. At least there should be any option to disable the new behaviour. Upstream even has not commented this bug in months. As you can see the latest release was just planned as a security release after 2 years there is not much development in rsync anymore.
Is there any decission on this subject. I've reinstalled the mga5 package which fxied the issue for me. But this has to be fixed for future versions. Should I post this to upstream? This bug breaks numerous scripts.
@David: I'm not sure if there is much development/maintenance going on upstream. I've posted upstream, but there is no reaction. Even it looks like there is only one man maintaining this package. One thing is sure, the updated package has changed its behavior and there are good reasons to keep the old behavior: file-lists of changed files/directories are generated by scripts and should process after some time (e.g. 30s). If the user has deleted the local directory in between, the old rsync continues while the new one just fails). Since there is always time between passing the file list and the execution of rsync there is always a chance for the new rsync to fail. As the man page states this is the idea of --ignore-missing-args and --ignore-errors. It should continue, not fail in case of errors.
I can do the patching, and it would not have effect on others (as you stated missing dirs), since you have to specify the given options to do so.
The problem is "solved" upstream. There is an option "--no-implied-dirs" which solves this problem. I'm not sure, if really everyone is aware of this option.
Resolution: (none) => WONTFIXStatus: NEW => RESOLVED
I have to reopen this issue. By test it looks like the behaviour still is changed!
Status: RESOLVED => REOPENEDResolution: WONTFIX => (none)
this is still a big issue for me, can we please change back to the old behaviour? There is NO workaround!
Priority: Normal => High
Assignee: pkg-bugs => mageia
Fixed in cauldron.
Version: 6 => Cauldron
fxied in cauldron
Status: REOPENED => RESOLVEDResolution: (none) => FIXED