Date: Fri, 16 Apr 1999 07:52:47 -0700 (PDT) From: Super-User <root@danberlin.resnet.rochester.edu> To: eagle <eagle@eagle.phc.igs.net> Cc: Amancio Hasty <hasty@rah.star-gate.com>, Gary Schrock <root@eyelab.psy.msu.edu>, freebsd-hackers@FreeBSD.ORG, docs@FreeBSD.ORG Subject: Re: NT4 server 2.5 times faster than Linux Message-ID: <Pine.SGI.3.96.990416074505.52171A-100000@danberlin.resnet.rochester.edu> In-Reply-To: <Pine.BSF.3.96.990415194126.85605A-100000@eagle.phc.igs.net>
next in thread | previous in thread | raw e-mail | index | archive | help
Wouldn't have mattered in this case. The samba team's beef is that Mindcraft is that they changed the widelinks setting. What widelinks does is check to see where a symbolic link is going. Without it, if i put a symlink to / or whatever in the wrong dir, well then that person now has *full* access to my puter. As you would imagine, any smart samba admin changes the setting from the default, to avoid this (or has no symlinks anywhere on the system). You take a *large* performance hit though, as it adds a bunch of various calls. IMHO, this is the fault of the samba team. They claim it takes 6 extra chdir and such calls for every file access. I can't see the reason you'd need that many, or that they don't do something to alleviate the problem of checking every single link every single time. There was also a kernel bug in 2.2.2 that was fixed in 2.2.5 that would have a major impact on performance. Of course, the linux advocates don't seem to get the fact that just becuase your kernel is buggy, doesn't mean someone else screwed up in testing with it, since it was what was available at the time. You know, the whole "actually getting real work done" thing? Mindcraft also posted to a few usenet newsgroups asking why they were getting such abyssmal performance, and the tradition of the great helping linux community, they got no answer. I could understand complaints if they got the answer "The kernel you are using has a bug in it, here's a patch, or wait for a fix, or else you'll take a major performance hit". But they got no such answer. "Why didn't they use 2.0.36" i hear also. Well, because everyone and their grandmother said to upgrade to 2.2.0. After all, it is in the stable series, right? It should be, ya know, stable? Although Mindcraft's past tests make me weary, in this particular case, i don't see how they screwed up all that badly. --Dan On Thu, 15 Apr 1999, eagle wrote: > > > > CC'd docs > On Thu, 15 Apr 1999, Amancio Hasty wrote: > > > > > I am trying to suggest to plug the hole that got Linux into trouble : > > > > Tuning configuration documentation, sample harware configuration > > and software configuration for a high performance web server and > > preferred mailing list to contact for performance or tuning questions. > > > > The later is important for in the past people have posted on usenet > > and their queries have gone unanswered or worse they > > didn't know where to send their questions to . > > > > > This sounds like a great idea, freebsd tuning docs ... > > > rob > > > > To Unsubscribe: send mail to majordomo@FreeBSD.org > with "unsubscribe freebsd-hackers" in the body of the message > To Unsubscribe: send mail to majordomo@FreeBSD.org with "unsubscribe freebsd-hackers" in the body of the message
Want to link to this message? Use this URL: <https://mail-archive.FreeBSD.org/cgi/mid.cgi?Pine.SGI.3.96.990416074505.52171A-100000>