Sync even on errors?

So, moving my rsync’d server backups to acd, I get stuff like below. (I rsync to a backup drive, then rclone those to the cloud. The idea is that it should keep things from changing while backing up to the cloud, and keeps 2 copies win/win) So some reason, I only get this from my PI backup, my normal server doesn’t get these errors for some reason.

2017/03/03 00:03:49 ERROR : etc/mtab: corrupted on transfer: sizes differ 0 vs 2815
2017/03/03 00:03:51 NOTICE: var/agentx/master: Can’t transfer non file/directory
2017/03/03 00:04:46 NOTICE: var/run/minissdpd.sock: Can’t transfer non file/directory
2017/03/03 00:04:46 NOTICE: var/run/acpid.socket: Can’t transfer non file/directory
2017/03/03 00:04:46 NOTICE: var/run/rpcbind.sock: Can’t transfer non file/directory
2017/03/03 00:04:46 NOTICE: var/run/initctl: Can’t transfer non file/directory
2017/03/03 00:05:34 NOTICE: var/run/NetworkManager/private-dhcp: Can’t transfer non file/directory
2017/03/03 00:05:34 NOTICE: var/run/NetworkManager/private: Can’t transfer non file/directory
2017/03/03 00:27:41 ERROR : var/run/atop/atop.acct: corrupted on transfer: sizes differ 128960 vs 129472

I don’t think the notices will stop a sync, but the error’s will. Is there a way to cause a sync to give you a list at the end of errors, and ask if you want to sync anyway? That way I can keep my backups somewhat clean. The errors I think are not real errors per say, just things rclone doesn’t like system file wise. Or am I reading this wrong?

and what exactly is it finding that isn’t a file or dir?

You shouldn’t be backing up anything in /var/run. Those are sockets and pipes and other stuff that gets created on boot. No need for recovery of Linux. /var/run is linked to /run. Just exclude it. I realize this isn’t you’re question but figured i’d mention that.

–exclude=/dev/
–exclude=/proc/

–exclude=/sys/
–exclude=/tmp/

–exclude=/run/*\

For the errors, the file is changing while you’re copying. You may be able to get away with adding hte flag --ignore-checksum Skip post copy check of checksums.

The ERRORs won’t stop the sync, it just prevents that file transferring, and causes a retry.

Either exclude the files like @calisro said, or you might consider using the --one-file-system flag provided you only actually have one disk.

By stop the sync, I meant preventing the delete at the end. It doesn’t delete because there were errors. I’ll look into the --one-file-system flag. Mainly, this won’t be much of a workable backup anyway, since permissions and whatnot will be borked. But I wan’t as much of a copy as possible, just in case I need something.

What exactly does the --one-file-system flag do, and how would it help me here?

I had excluded the /run from the rsync this is copied from (I’m sending 3rd copy to acd). But I’ll add var/run as well to my exclude file, good to know!

As for the size issues, couldn’t have been changing. This is a backup, not the live version… I’ll investigate more.

You can use --delete-during if you want stuff to be deleted as it goes along, not wait until the end

That will skip the pseudo file systems like /proc /sys /dev