On Friday 11 October 2002 08:19 am, Jason Straight wrote: > On Thursday 10 October 2002 07:17 pm, Todd Lyons wrote: >> Jason Straight wrote on Thu, Oct 10, 2002 at 09:45:45AM -0400 : >>> If I had a nickel for every file reiserfs fragged on a busy server it >>> still woudln't come close to paying for the downtime - ext3 is rock >>> solid. >> Reiser definitely has its benefits. I have seen reports of people that >> use it for news servers where expires normally take hours. With reiser >> it does it in (IIRC) 15 minutes or less. When things go wrong, a newbie >> is scared by the daunting task of repairing the fs.
>> [...] he swore to never use Reiser again. (That's the wrong >> attitude because on any other file system, the abuse he put it through >> might have rendered it completely useless). > For me the problem was that files which weren't even being written to would > get fragged with data from other files ending up mixed in with them. This > would happen on machines that had been running for long periods of time > with uptimes nearing 200+ days. Suddenly files like sendmail.cf, > httpd.conf, and probably many others I didn't notice on multiple servers. I've just had a power failure at a client site and Reiser killed a file in /usr/lib/ somewhere needed for PostFix: when I try to start PostFix, that process and anything that tries to go through the filesystem after that (on that machine, anything except Pick/D3, and even that when it does a sync() call) freezes. Cheers; Leon