# chmod a-r /usr/lib32/libpulse{-simple.so.0.0.2,.so.0.8.0,common-0.9.15.so}

source: http://wiki.debian.org/skype

- dns authoritative name server being development, with serial version and difference between them
- puppet and func integration on roadmap
- online monitoring
- graph and monitoring perhaps

http://video.linuxfoundation.org/categories/2009-linux-plumbers-conference

raiser4, ext3 and ext4 using iozone

it seems a good i/o benchmark, it’s being used by ext4 and btrfs developers

I saw some interesting sessions in the schedule:

• Managing KVM guests with Common Information Model
• DRDB 8.3 and Beyond
• Merging KGDB, KDB and Kernel Mode Setting
• some sessions talking about threads

It’s going to start in September 23 to 25, and it takes place in Portland, Oregon

http://0pointer.de/blog/projects/mutrace.html

$ LD_PRELOAD=~/projects/mutrace/.libs/libmutrace.so epiphany

[...]

Mutex #223 (0x0xbbd4f0) first referenced by:
/home/wfelipe/projects/mutrace/.libs/libmutrace.so(pthread_mutex_init+0x140) [0x7f3630ff14b0]
/usr/lib/libnspr4.so.0d(PR_NewLock+0x34) [0x7f36309c0aa4]
/usr/lib/libnspr4.so.0d(PR_CreateStack+0x56) [0x7f36309b3446]
/usr/lib/libnspr4.so.0d [0x7f36309aa1fb]
/usr/lib/libnspr4.so.0d [0x7f36309c5930]
/usr/lib/libnspr4.so.0d [0x7f36309b841d]
/usr/lib/libnspr4.so.0d(PR_NewLogModule+0x1a5) [0x7f36309adb65]
/usr/lib/xulrunner-1.9/libxul.so [0x7f3622a3db62]
/usr/lib/xulrunner-1.9/libxul.so [0x7f3622a5c3a6]

mutrace: Showing 10 most contended mutexes:

Mutex #   Locked  Changed    Cont. tot.Time[ms] avg.Time[ms] max.Time[ms] Flag
404      433      390        9        1.291        0.003        0.011 x.a-
38        9        6        1        0.082        0.009        0.035 x.a-
208        7        2        1        0.054        0.008        0.048 x.a-
423        7        2        1        0.036        0.005        0.017 x.a-
212       91       44        0        0.189        0.002        0.023 x.a-
400     1181       20        0        0.594        0.001        0.004 x.a-
396      727       17        0        1.059        0.001        0.160 x.a-
518       18       17        0        0.012        0.001        0.001 -.a-
397      265       16        0        7.724        0.029        5.708 x.a-
223      811       14        0        0.371        0.000        0.001 -.a-
...      ...      ...      ...          ...          ...          ... ||||
/|||
State:                          x = dead, ! = inconsistent /||
Use:                          R = used in realtime thread /|
Type:          r = RECURSIVE, e = ERRRORCHECK, a = ADAPTIVE /
Protocol:                               i = INHERIT, p = PROTECT

mutrace: Note that the flags column R is only valid in --track-rt mode!

mutrace: Total runtime is 8136.226 ms.

mutrace: WARNING: 43 internal hash collisions detected. Results might not be as reliable as they could be.
mutrace:          Try to increase --hash-size=, which is currently at 3371.

sources: http://git.0pointer.de/?p=mutrace.git

I finally checked the most spent syscall in CPU in both web accelerators:

varnish

% time seconds usecs/call calls errorssyscall
—— ———– ———– ——— ——— —————-
60.79 35.816504 224 159674 futex
19.52 11.498331 11067 1039 poll
17.12 10.085451 672363 15 nanosleep
2.28 1.343794 671897 2 restart_syscall

squid

% time seconds usecs/call calls errors syscall
—— ———– ———– ——— ——— —————-
99.98 1114.225617 747 1490839 112207 futex

it seems squid hasn’t been built for performance in stressful environments :/

http://github.com/wfelipe/personal/blob/master/programming/inotify.c

Online updates, filtering files, exactly changes, a wide open watcher :o)