File Systems

It turns out that we have to optimize our FS before benchmarking. I have read some articles and maillist archives so I got some impression:

1. Give correct parameters to newfs(1)
   avgfilesize, filesperdir, will control DIRPERF hashing.

2. Turn on noatime mounts.

3. etc.

Just make a note.

Busy @ work for coda afs and nfsp for FreeBSD…

For the Simplified Chinese project: We are having clean-ups for the handbook, it is very nasty as the format is changed unwillingly by the translators…

Finally I am thinking about porting ReiserFS4 to FreeBSD. The current ReiserFS is GPL’ed so it should not get into FreeBSD kernel, but I think it might be OK to leave it as a ports of kernel module (like pf).

While reading the snapshot code in FFS Soft Updates’s FreeBSD implementation, I got a question about the following code:

 1
 2
 3
 4
 5
 6
 7
 8
 9
10
11
12
13
14

		/*
		 * If there is a fragment, clear it here.
		 */
		blkno = 0;
		loc = howmany(xp->i_size, fs->fs_bsize) - 1;
		if (loc < NDADDR) {
			__len = fragroundup(fs, blkoff(fs, xp->i\_size));__
			if (len < fs->fs_bsize) {
				ffs_blkfree(copy_fs, vp, DIP(xp, i_db[loc]),
				    len, xp->i_number);
				blkno = DIP(xp, i_db[loc]);
				DIP_SET(xp, i_db[loc], 0);
			}
		}

Here:
还是让我来诠释一下吧：

1.决不引进新的文件系统，除非有人能证明我们现有的文件系统无法存储某种文件.

2.看清谁是我们的用户和看清楚谁不是我们的用户一样重要,我们要让我们的用户时刻"充满希望",我们的用户都是乐观向上地,他们只要能看到 将来 我们有可能 在现有基础架构上 实现 和 现在 别人的用户 正在 使用的功能 的希望 就满足了.

3.我们决不会闭门造车,你看,我们要统一"raid frame"的 雄心 就是 被 那个 一直饱受(翻译的话请使用过去将来进行时)“Adaptec SCSI Host raid"折磨的用户 所激励地.

4.如果我们 能 建议用户不加载ACPI模块 并 能让那些 被那个蹩脚的ACPI奴役的 用户们 接受 我们的建议 的 话, 那么我们也许最好干脆不去解决现在的ACPI问题了, 因为我们 需要 再次地 去 完整地 理解 一下新版本的ACPI呀.

5.请看,我们只用了10%的努力去维护几个论坛和邮件列表 就 出色地解决了每个RELEASE版本中90%的默认配制下的BUG, 这是个更多么简单的解决方案啊.

6.我们用PORTS把用户从复杂的第三方软件安装操作中隔离并解救出来,用户只需要简单地make install就可以啦….什么?你想在编译之前加一些配制参数?你完全可以把我们映射过的参数放到环境变量中啊…..什么?我们映射出来的参数不全…这可真是遗憾啊,不好意思啦,没有参考文档,还是请您自己研究一下Makefile文件吧,我们相信你的实力,您是完全可以自己搞明白Makefile文件中的那些变量的.

7.我们提供了很多的功能和机制以及大量的参数,我们相信您可以根据您的实际应用总结出一套配制方案.这样对您是很有好处的,因为通过这个配制过程,您可以总结出一套适合自己的策略,这样您所实现地应用就会比那些不了解您的"策略"的人所实现的应用更加出色,而且您还可以比别人"更会"使用我们的XxxxXXX.

主要是空格的问题。我打算接下来的一段时间琢磨琢磨SoftUpdates，如果这个patch能commit的话，delphijfork的diff一下就能减少46K :-)

Sprite is a research operating system developed at the University of California, Berkeley, by John Ousterhout’s research group.

Sprite is a distributed operating system that provides a single system image to a cluster of workstations. It provides very high file system performance through client and server caching. It has process migration to take advantage of idle machines. It was used as a testbed for research in log-structured file systems, striped file systems, crash recovery, and RAID file systems, among other things.

See the original letter here.

I personally manages several FreeBSD 5.2-CURRENT instances, and here are my experiences.

FreeBSD 5.x contained many new features, technologies. Currently, it’s clear that 5.2-CURRENT is more stable and have a better performance than 5.2(.1)-RELEASE. Therefore, if you are looking to install a new (of course, and ONLY if you choose 5.x series!) computer, the best solution would be 5-CURRENT.

I finally understood what he want, according to his post today:

" Hey, here’s an idea… using the VFS wrapper idea our
‘ports’ system could create a namespace that makes everything
appear to be preinstalled. So instead of CDing somewhere and
typing make, or running an apt-get type of utility, you would
simply run the program you want to run and our package manager
would download and install it on the fly."