On Wed, Jun 10, 2020 at 8:41 AM Josh Fisher <jfis...@jaybus.com> wrote:
> I still feel that Bacula's design is correct. Yes, 802.3az changes the > always-on nature of a connection, allowing either side to temporarily > power down its transmitter to save energy, but the standard itself > doesn't change the original goal of a persistent connection. It is the > switch firmware and/or NIC device drivers that claim to support it, but > do not. It makes sense for Bacula to be as robust as possible, but this > is not a Bacula design flaw. It is a work-around for buggy hardware. > I've also run into this when trying to back up over a VPN. The backup time can easily exceed the VPN's maximum session time. It's fair to argue that both NAT routers and VPNs are a corruption of TCP/IP's design intent, but it doesn't seem likely we'll be rid of them any time soon. Bacula doesn't work very well with either. Besides the connection drop issues, I haven't yet gotten client-initiated backups to work from behind a NAT, and I haven't found anyone who's confirmed they have it working, either. None of this, of course, is an issue when backing up always-on servers with static IPs -- which is Bacula's focus. The problems come in when it's used to back up endpoints. Unfortunately I haven't yet found anything else to use for that that lets me control my own data and isn't a subscription model. (In academic departments, it's much easier to find money for one-time expenses than it is to find a consistent source of it.) -- David Brodbeck System Administrator, Department of Mathematics University of California, Santa Barbara
_______________________________________________ Bacula-users mailing list Bacula-users@lists.sourceforge.net https://lists.sourceforge.net/lists/listinfo/bacula-users
