From: Peter G. <gr...@gr...> - 2004-11-11 20:00:53
|
Hello, My backup "suddenly" died a week ago. Familiar story for developers, and emails starting like that make that funny chilly feeling of "poor bastard, he didn't realise that unexplained random errors are close to impossible to fix". Good news, I got lucky, I believe at least I know the problem, and avoided it. :) I reconfigured the whole mess and (among other uninteresting changes) included "--compress" in my rsyncd process' param list. The results were catastrophic: backups that took 200 minutes usually died in 200-400 minutes by "received signal=ALRM" or "Child exited prematurely" with no other visible reason (it would be neat to have a bit explanation about these errors in logfile for the inexperienced or clueless, translating it to "timeout on server/on client/local rsync probably died/remote rsync probably died/etc"). Zero backups were done on several machines. I tried various voodoo tricks and finally removed --compress, and things became shiny happy again. "Is this the way it has always been? Could it ever have been different?" -- Pink Floyd Thanks, Peter |