电脑不够用了,上周末去苹果家园买了台 MBP 990,这是目前已上市最新款中最低配的 MBP,标配 2G 内存,购机时给升到了 4G,内存有些贵(650元),不过用惯了 4G 内存的本本,不想再倒回到一年多以前,用了几天发现,升级到 4G 内存实在是一个非常明知的选择,并且 4G 内存用 Snow Leopard 也实在不算多。
使用Nginx一分钟搭建全功能的Twitter/Twitter Search API
现在有很多的 Twitter 代理软件,比如 tweetr 和 [birdnest](http://code.google.com/p/birdnest/, 不过使用反向代理来做的话会更容易,并且得到的是一个 全功能 的 Twitter 代理,这是我在我的 VPS 上使用 Nginx 配置 Twitter 和 Twitter search API 代理的示例配置文件(当然了,这里写出来的域名是假的,哈哈):
修改 OpenLDAP 的 Base DN
在 Debain 里安装 OpenLDAP 时,Debian 会提示给 LDAP 的 admin 用户设置一个密码,然后就自动地创建了一个默认的数据库,这个默认的数据库使用了一个默认的 base DN,默认情况下,Debian 会使用本机的域名来作为 base DN,比如如果我的域名是 xupeng.local,那么 Debian 就会使用 dc=xupeng,dc=local 作为我的默认 base DN,但是很不幸,我没有给我的 Debian 测试机设置域名,于是我的 LDAP 默认数据库的 base DN 就成了 dc=nodomain,要想修改 base DN 还真不是直接改了配置文件(/etc/ldap/slapd.conf)就行的,不过 Google 一下很快就找到了解决方法:
Linux 进程的 Uninterruptible sleep(D) 状态
运行在 KVM 虚拟机里的一些进程突然出了问题,这些出了问题的进程无法用 kill 杀掉,使用 ps 可以看到这些进程处于D 状态:
Settings for Improving Poor 3D Performance of Intel 945GM
I don’t use compiz on my work computer(HP 520 laptop), but I do use compiz on my own laptop which has a NVIDIA 8400GS video card, the main reason I don’t use it on the HP 520 is that: it’s really too slow to run compiz smoothly with the default settings of Xorg and intel driver, but I hear that the 3D performance has been greatly improved in kernel 2.6.29, so I spent some time to get compiz work well with the HP 520(Intel 945GM video card).
Sun 被 Oralce 收购,Virtualbox 前途未卜
Sun 被 Oralce 收购了,我担心我没有新版本的 Virtualbox 用了……
Virtualbox 是我最喜欢的桌面虚拟化软件,目前的最新 release 是 2.2.0。我之所以喜欢 Virtualbox,是因为 Virtualbox 实在是:
VMWare ESXi 3.5.0-153875 on HP 520 Laptop: “PCPU 1 Locked Up…”
But I was not surprised at all, because I knew there were problems with the HP 520, at least with this model of HP 520. This HP 520 is equipped with Intel Core 2 T5200 CPU, according to wikipedia(http://en.wikipedia.org/wiki/List_of_Intel_Core_2_microprocessors), it looks like this CPU is an OEM product. It’s a Core Duo 2 CPU, but it can’t run 64bit OS, and no hardware virtualization support either, really weird.
Anyway I can’t get it work with VMware ESXi, a little frustrated, I have to use KVM VMs for testing krugle before the new server reaches.
Gnome-do 0.8.1.3 Doesn’t Show Application Items
The problem with my gnome-do 0.8.1.3 is: it doesn’t list anything of my application items(The ones sits in Gnome’s Applications menu), I think this functionality is of gnome-do itself. But all the plugins of gnome-do work well, it’s really weird why it breaks while 0.8.0 worked for me.
Window Doesn’t Redraw Correctly When Using Compiz
Window doesn’t redraw correctly sometimes when using compiz, but if I switch to another window then switch back, it redraws, this problem occurs with any type of windows, it’s annoying.
I am using compiz 0.8.2 on my DELL 1420, with a Nvidia 8400M GS video card and Nvidia driver of version 180.44. Once upon a time, I can run compiz on this computer without any problem, but that’s really a very long time ago, with a very old nvidia driver and very old compiz fusion.
After looking around the web, I found there are people with the very same problem as mine, some of them have same video card as mine too, and there are several solutions to this problem, but maybe one of the methods can solve one guy’s problem, but it doesn’t matter to other people, I have tried some of them, but none of them matters, following are some methods maybe useful(at least useful for some people to get this problem solved):
- Use “Indirect Rendering” option to start compiz
- Use “Loosing Binding” option
- Put in ~/.gnomerc this line :
dia-settings -a InitialPixmapPlacement=1 -a GlyphCache=1 - The problem is caused by an Nvidia feature called “powermizer” that pushes the video card into low performance mode, so load nvidia kernel module with these parameters:
1 2 | |
I’ll try these solutions one by one, and maybe combination of these solutions, hope this annoying problem can be fixed.
UPDATE: It looks like the problem has gone after I enabled “Indirect Rendering” and “Loosing Binding” at the same time, wow!
Unmount the Unreachable NFS Volume
I previously mounted one NFS volume from the NFS server, but now the NFS server is down, then I have problem with “df” command to see the disk space usage information from my computer, now the only thing I want to do is to unmount the unreachable NFS volume.
The NFS volume is mounted at /mnt/nfs, when I try to unmount the NFS volume, I am told that it is busy, so I use lsof /mnt/nfs to find out the process using it normally, but lsof gets stuck and it never return the shell prompt to me. Ok, then I use fuser -9 -k /mnt/nfs, try to kill the process forcefully, but fuser gets stuck too…
All methods I normally use to unmount a volume fail, so I go back to RTFM, and I get two interesting options of umount:
1 2 | |
Good, it looks like what I need, but unfortunely I fail with umount -f /mnt/nfs all the same, so I move on to another option:
1 2 3 | |
Wow! It works like a charm! That’s it: “umount -l /mnt/nfs”, maybe it will never be a pain to face a unreachable NFS volume again.