From: Les M. <les...@gm...> - 2005-09-16 13:02:00
|
On Fri, 2005-09-16 at 02:16, Peter Padberg wrote: > > > But the problem is, when a customer lost data, I need the backup fast > > > and I have no time to wait "some" hours. > > > > As I said above, you don't have to wait. > Ahh come on, why Backuppc offers a webtool? Command lines nearly always offer more options than their web and GUI simplified fill-in-the-form counterparts. > Why I can download restorefiles immediately, > but must wait to restore files to servers, some hours!? > Sorry, but there is no context. The difference is that backuppc does most operations through a scheduling server, and generally the web interface simply passes commands to the server which schedules the operation. The direct download is an exception, because the web server must do that immediately to pass it back on the same connection. The scheduling operations are generally controlled by settings in config.pl, but the nightly run is an exception that blocks everything else - it removes files in the pool that no longer have any other links and might accidentally remove files being added if backups were allowed to run at the same time. That said, I'm not sure why you couldn't do a restore at the same time. > > The web interface does not do everything you can > > do through the command line. > And when some external customers want to restore files, I must login > into as root to realize it!? You really only need to be the backuppc user to kill and restart the process, although if you use something like the RedHat init script and the 'service backuppc restart' command you might need to be root. > Then I hope the next time, it is not saturday night. I assume that means you don't trust your saturday night person to do root operations. You can always configure sudo to permit anyone you choose to run a restart script. > I know some other backup-software, where I can restore files without any > hours to wait. :-/ That software doesn't save your disk space by linking duplicate files. The reason for the nightly run blocking everything else is to protect those links. > > There are settings that may reduce the time used by the > > nightly run. > > See the options for $Conf{MaxBackupPCNightlyJobs} > > and > > $Conf{BackupPCNightlyPeriod}. > Yes, of course. > And when I need a restore in this time? Set up web logins for everyone to get their own files through the web interface without bothering you or the operator you don't trust. > My question to development: > Why I can download restorefiles immediately, > but must wait to restore files to servers, some hours!? I think you do have a point, even though there are workarounds. I asked the same question long ago and Craig pointed out the race condition while adding new files but I don't see why it would apply to a restore. Couldn't the scheduler let a restore start anyway? -- Les Mikesell le...@fu... |