New issue
Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.
By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.
Already on GitHub? Sign in to your account
Regression: When run through BackupPC_dump, 3.2.3 dumps file contents in log #95
Comments
I'm the BackupPC developer. BackupPC 3.x is very old, and it's quite possible it's broken with the latest rsync. BackupPC 4.x uses a completely different integration with rsync, and it much more up-to-date with recent and current rsync versions. Let's work on this issue using the filed BackupPC issue, and we can close or update this issue based on what we find. |
Well either way, doesn’t matter at this moment because my computer is fried. So I won’t be screwing up your projects anymore.
Robert J Collins
|
After checking into this, this change does indeed break BackupPC 3.x. We confirmed that specifying BackupPC 3.x only uses rsync protocol 28. Would it be possible to make |
Nobody has specified what the actual rsync issue is. Completely remove all mention of BackupPC and specify what error behavior is happening. Is a v3.2.3 rsync being run as a remote server and the local process doesn't have a stderr open for the remote rsync to be using? |
Yes, rsync <-> rsync works fine. This is due to a long-standing bug in the perl module I just released a new version 0.76 of Please go ahead and close this issue. |
IMO, rsync should not enable msgs2stderr by default, at least for rsync protocol 28. Even tho this is not an rsync bug, it will affect users of older It's your call tho, @WayneD |
I'm a bit torn on this one. I've currently got a potential change in the rsync-patches repo (stderr-compat.diff) that you could try out if you like. |
I tried the patch and confirmed it works fine with both the old (broken) It would be great to include this in the next rsync version, since It is completely up to you - I'm fine with telling BackupPC 3.x users that encounter this issue to either upgrade to BackupPC 4.x (which does not use Thanks for all you do leading rsync development! |
Hello, I know nobody is supposed to install unstrusted RPMs from Internet, but in case someone trust me and want to do this anyway, here is a zip (because GitHub) containing RHEL7/CentOS 7 SRPM/RPMS of the fixed perl-File-RsyncP package (for the ones having CentOS 7 BackupPC servers): perl-File-RsyncP-0.76-1.el7.zip Just deployed on tested on my servers, confirmed fixing the issue ! Regards, Adam. |
FYI, this was reported on Debian against rsync: https://bugs.debian.org/cgi-bin/bugreport.cgi?bug=969463 But from what I understand, this is actually a bug in libfile-rsyncp-perl, which is no longer present in Debian testing and stable, so I think it has very little chance of seeing an update: https://packages.debian.org/search?searchon=sourcenames&keywords=libfile-rsyncp-perl |
I've gone ahead an merged in the fix that I worked up in rsync-patches. |
This makes the default for a protocol-28 server process be --stderr=client instead of --stderr=errors. See rsync's github issue #95.
Hello,
592059c has introduced a bug in combination with BackupPC 3. Dumps work fine until a vanished file is encountered, at which point file data is dumped directly into the log, breaking the backup process.
For reference, in my use case, BackupPC calls rsync on the client device via ssh and this cmdline:
/usr/bin/rsync --server --sender --numeric-ids --perms --owner --group -D --links --hard-links --times --block-size=2048 --recursive --checksum-seed=32761 --ignore-times . /
I don't understand rsync quite well enough to tell if it's a BackupPC or rsync issue. However, due to the vast number of BackupPC installs out there that will not receive timely (or any) updates, but still have to backup clients which will receive rsync updates, I'd say that not having such a breaking change in rsync is preferable.
On top of this, it's very difficult to figure out why this breaks BackupPC. It's possible it also breaks other users of rsync. And while
--no-msgs2stderr
can be used to avoid this bug, older rsyncs won't recognize it and abort, prohibiting its widespread use.Also see backuppc/backuppc#369
Thanks!
The text was updated successfully, but these errors were encountered: