hns - 日記自動生成システム - Version 2.19.9

先月 2010年04月 来月
1 2 3
4 5 6 7 8 9 10
11 12 13 14 15 16 17
18 19 20 21 22 23 24
25 26 27 28 29 30
Namazu for hns による簡易全文検索
詳しくは 詳細指定/ヘルプを参照して下さい
検索式:

2010年04月20日(火) 旧暦 [n年日記] [更新:"2010/04/23 11:33:29"]

#1 [機械] raidctl -R /dev/wd0h raid1 を手で実行

NetBSD tera5 4.99.3 NetBSD 4.99.3 (KUROBOX.raid) #1: 
Wed Jun 13 01:01:52 JST 2007  
root@trapani:/work/20061019/sys/arch/evbppc/compile/obj/KUROBOX.raid evbppc
RAID5 の機械で、
ttyp0:root@tera5 9:16:20/100420(~)# raidctl -s raid1
Components:
           /dev/wd0h: failed
           /dev/wd1h: optimal
           /dev/wd2h: optimal
           /dev/wd3h: optimal
No spares.
/dev/wd0h status is: failed.  Skipping label.
このようになっている時に、この raid1 の disk の mount や exports を外してから、
/sbin/raidctl -R /dev/wd0h raid1
をしたら、 とても load average が高い状態で長い間かかって、 直った。これは mount が外せたから良かったが、/ になっている ものはこれが簡単には出来ないと思う。 別の起動 disk を使う必要があるのかな。

この機械は多分 netboot は出来ないと思うけれど、 もし netboot の出来る機械なら sysinst の入っている Memory Disk kernel に raidctl が入っていると嬉しいと思う。

raidctl 入の Memory Disk Kernel (MD Kernel):

2008/04/23/ の日記を見ると、 kuro-box/ に置いてある netbsd-KUROBOX_MD.bin には (sysinst と) raidctl が入っているらしい。 USB メモリを抜いておいて linux 側で login し /boot で
wget http://www.ki.nu/~makoto/kuro-box/netbsd-KUROBOX_MD.bin
すれば良いような気がして来た。
↑これがどうも間違いで、少なくとも次の日付のものには入っていないようだ (2010/04/22 追記)
-rwxr-xr-x  1 makoto  wheel   6078208 Apr  1  2008 netbsd-KUROBOX_MD.bin
root@catania:/export/src-4.0/sys/arch/sandpoint/compile/obj/KUROBOX_MD
NetBSD 4.0.0_PATCH (KUROBOX_MD) #3: Sun Mar 2 18:02:28 JST 2008

raidctl 入の Memory Disk kernel はこちら:

以前に入れて作ったものが、 作ったところにまだ置いてあったので以下に写した 版は上のものと同じで NetBSD 4.0.0_PATCH sandpoint
wget http://www.ki.nu/~makoto/kuro-box/netbsd-KUROBOX_MD-raidctl.bin
mkdir /usb-mem しておいて、/etc/fstab に次のような行を書いておく
/dev/sd0d       /usb-mem msdos  rw,noauto 0 0
その上で、手で mount /usb-mem し、 その /usb-mem に入っている netbsdboot を調整することにより、MD kernel と通常 kernel の切替も可能 だと分った。(終ったら umount /usb-mem しておいた方が良い)


2010年04月19日(月) 旧暦 [n年日記] [更新:"2010/04/20 10:26:45"]

#1 [pkgsrc][NetBSD] pkgsrc/wip/scrc

pkgsrc/wip/scrc に入れられるようにと まずは手元で を作っている。i386 (5.99.27) と amd64(5.0_STABLE) で試している
PKG_DESTDIR_SUPPORT= はまだ出来ていない
それで、何故か:
  • amd64 の方で gcc3-c を必要とすると思い込んで作りに行く (そしてそれが作れないで失敗する)
    => Full dependency unix2dos: found unix2dos-1.3
    => Full dependency gcc3-c>=3.3.5nb1: NOT found
    => Verifying reinstall for ../../lang/gcc3-c
    ....
    
    Created "Makefile" in /export/pkgsrc/lang/gcc3-c/work/gcc-3.3.6
    exec: /usr/pkg/gcc34/bin/gcc: not found
    
    これも変な話だ、
    • rm -rf work しても
    • /etc/mk.conf を消しても
    • scrc2 を全部消して cvs update -dPA しても
    • root になっても
    こうなってしまう。
    modena# which gcc
    /usr/bin/gcc
    modena# gcc --version
    exec: /usr/pkg/gcc34/bin/gcc: not found
    modena# 
    
    あれ、置替っている ?
    modena# ls -l /usr/bin/gcc
    -r-xr-xr-x  2 root  wheel  43 Apr 16 08:55 /usr/bin/gcc
    modena# cat /usr/bin/gcc
    #!/bin/sh
    exec /usr/pkg/gcc34/bin/gcc "$@"
    modena# 
    
    この二つは comp.tgz から戻しました
  • amd64 では、一度は作れたことがあって、しかしその時に
    modena@makoto 17:03:24/100419(..personal-pkgsrc/scrc2)%  \
    /usr/pkg/scrc-2.1/bin/scrc --help
    Shared object "libCcDriver.so" not found
    
    などとなって、これが解決出来なかった (その内に前項の問題になってしまった)
  • i386 の方では
    <stdout>:1168:31: error: macro "PRS_wrap" passed 1 arguments, but takes just 0
    PRS_Scanner.h:128: error: conflicting declaration 'unsigned int PRS_lineno'
    
    のようなことを言って止ってしまう

gcc3-c を必要とすると思い込んで ..:

そう思い込んでいる記録 (make -dA で表示される膨大な譜) の中に
ParseReadLine (255): '_GCC_PKGSRCDIR=           ../../lang/gcc3-c'
Global:_GCC_PKGSRCDIR = ../../lang/gcc3-c
ParseReadLine (256): '_GCC_DEPENDENCY=  gcc3-c>=${_GCC_REQD}:../../lang/gcc3-c'
のような部分があったので、次のように grep して見た
modena@makoto 01:35:52/100420(..personal-pkgsrc/scrc2)% 
sudo grep -r 'gcc3-c>=3.3.5nb'  /usr/pkg /export/pkg rc/
/export/pkgsrc/lang/gcc3-c/buildlink3.mk:
	BUILDLINK_ABI_DEPENDS.gcc3-c+= gcc3-c>=3.3.5nb1
/export/pkgsrc/personal-pkgsrc/scrc2/work/.depends:
        full 	gcc3-c>=3.3.5nb1        ../../lang/gcc3-c
記録のその少し前に
ParseReadLine (249): '_GCC_PKGBASE=             gcc3-c'
Global:_GCC_PKGBASE = gcc3-c
がある。この(数字) は mk/compiler/gcc.mk の行数らしい。 面倒なので Makefile に次の行を加えたら、とりあえず、回避
+_GCC_REQD=             4
(あり得ない回避方法)


2010年04月18日() 旧暦 [n年日記] [更新:"2010/04/19 01:26:31"]

#1 [荷物] W-ZERO3 WS007SH が届いた

最近 Twitter で知合いになった人から、 家人のこわれた WS007SH の代替になる Willcom 電話機が届いた。
その後知合いから(貸してくれるという) IC-745 も届いた

#2 [無題] 榎木を移植

小さな庭の石畳の間に(結構)大きく高さ 3m、幹 5cm くらいに育ってしまった榎木と思われる木を移植した。
まず、石を一つづつ外した。どうもその石畳に、元々 4cm くらいの隙間が確保してあったようだ。 これを 1999 年に作った時に既にいくらかの太さがあったと思われる。
それで、その石を外した後に、 まわりを堀ってから、チェーンブロックで少し上に引いたら、 どうもそのまま抜けそうなので、 そのまま続けて引張ったら、土を全く付けずに抜けてしまった。
そうそう、その前に、このチェーンブロックを支える 3m くらいの三脚がある訳だが、これがこのせまいところで (ちょっと腰の具合も変な今日に) 立てられるかなと思ったが、一応何とかそれはうまく行った。
車の上に載せて固定し、 時速 60km は風速 16m/s かなと思いながら、それよりも速い走行で移動。
これで根付くかな、と思うが、考えたら、 以前に枇杷の木を移した時にも、土は殆んど付いていなくて、 それでも何とか根付いている気がする。
なお、きょうは写真はない。あまり時間がなかったのが正直なところ。
2009/05/24/ 2009/06/13/ いまはこのように茂ってはいない

#3 [WWW] westkit plug-in for Wordpress

中国語(?)なので良く分らないが、WordPress のプラグインで、 ソースに色付けしてくれると見た 実は 武者さんの WordPress がきれいなので、これかなと思ったという訳。Plug-in でなくて Theme なのかな。

#4 [NetBSD] mount_union

HEAD/i386/201004120000Z-obj/ な XEN3_DOMU で、
rm -W patches/patch-aa すると、かなりの確率で panic
panic: kernel diagnostic assertion "rw_lock_held(&wl->wl_rwlock)" failed: 
file "/home/builds/ab/HEAD/src/sys/kern/vfs_wapbl.c", line 1550
cpu0: Begin traceback...
_prop_dictionary_keysym32_pool(0,0,0,100400,1,800,800,8e06,0,cb6c4000) at 0xc10f1000
cpu0: End traceback...
WAPBL と union mount は、良く検証されていないのかな。 上と下の両方に何らかの処理が要るのに、その片方しかやっていない ? (というようなことは上の表示からは何も読めない)

Xen の console:

xm console 名前|番号
で卓が取れるが、これを重複して、他の窓からまた実行してしまうと、 何か変になる。気が付いて DOM0 で xenconsole の処理を kill すれば、 良いのだけれど、「卓は他で動いているので、取れないよ」と言ってくれ る方が嬉しいが、これは無理な相談かな。


2010年04月17日() 旧暦 [n年日記] [更新:"2010/04/19 00:07:05"]

#1 [Emacs] emacs22 と emacs23 の関数の違い


#2 [pkgsrc] SU_CMD= ${LOCALBASE}/bin/sudo /bin/sh -c

In It is written as:
9.12. Using 'sudo' with pkgsrc

When installing packages as non-root user and using the just-in-time su(1) feature of pkgsrc, it can become annoying to type in the root password for each required package installed. To avoid this, the sudo package can be used, which does password caching over a limited time. To use it, install sudo (either as binary package or from security/sudo) and then put the following into your mk.conf, somewhere after the definition of the LOCALBASE variable:

.if exists(${LOCALBASE}/bin/sudo)
SU_CMD=        ${LOCALBASE}/bin/sudo /bin/sh -c
.endif
But I could not confirm this. Am I still wrong or the document is obsolete ?
.if exists(/usr/pkg/bin/sudo)
SU_CMD= ${LOCALBASE}/bin/sudo /bin/sh -c
.endif
Now it works. I don't know really why.

Comments related this article

Re: SU_CMD= ${LOCALBASE}/bin/sudo /bin/sh -c by obache    2010/04/18 18:46
${LOCALBASE} が定義される前に mk.conf が読み込まれてるからですな。

2010年04月16日(金) 旧暦 [n年日記] [更新:"2010/04/19 00:07:29"]

#1 [EDA] personal-pkgsrc/scrc2 (scrc 2.1) もはまり中

scrc (SpecC Reference Compiler) 2.1 を動かそうとしていて、途中まで pkgsrc を作って見ているが いろいろと(自分で思うには)理不尽な反応があり、止っている。
  • Constant.cc:1177: error: `ubit2str' was not declared in this scope
    Constant.cc:1177: warning: unused variable 'ubit2str'
    Constant.cc:1191: error: `ubit2str' was not declared in this scope
    Constant.cc:1191: warning: unused variable 'ubit2str'
    
    scrc-2.1/src/IntRep/Constant.cc の次の行 が読まれていない ? というか別のものを見に行っている
        45  #include "bit.h"
    
  • 仮に、上に必要な次の二行を Constant.cc に手で加えておくと
    // Conversion to string
    char *bit2str(unsigned int base, char *endptr, const _bit& b);
    char *ubit2str(unsigned int base, char *endptr, const _bit& b);
    
    まだ _bit 関係の定義がおかしいようなことを言う。
Makefile に pre-configure: を使って、ここまでやっておいてから pkgsrc を使わないで gmake 出来るようにして見た
make configure
(cd work/scrc-2.1; env PATH=/usr/pkg/gcc34/bin:$PATH gmake)
が、やはり上の ubit2str は同じように言われるので、この問題は pkgsrc 由来でも、また gcc4 由来でもなさそうだ。 どれかのシンボリックリンクが悪さをしている気もする。
どうも __BIT がどこかで定義済ということらしく、 この名前を変えたら、 先に進んだ。

error: ambiguous overload for 'operator[]' in 'PrintBuffer..:

Expression.cc: In member function 
  `const char* SIR_Expression::Print(bool, unsigned int, int, int)':
Expression.cc:5173: error: ambiguous overload for 'operator[]' in 'PrintBuffer[LeadingNL]'
Expression.cc:5173: note: candidates are: operator[](const char*, long int) <built-in>
これ、もしかすると 64bit OS 非対応ということかな (long int <-> int)
Expression.cc:5173: note: candidates are: operator[](const char*, long int) <built-in>
/export/pkgsrc/personal-pkgsrc/scrc2/work/scrc-2.1/inc/GL_String.h:303: 
   note:                 char& GL_String::operator[](int)
どうもそうらしい、i386 で試したら、かなり先まで進んだ。
ERROR #3105: Semantic error in line 221
             in file "/usr/include/stdio.h":
             Declaration of symbol '__restrict'
             does not match former declaration:
             type mismatch (#2039).
やはり gcc34 必須かな。そうでもなさそう。gcc34 でも gcc 4.1.3(N) でも同じ。 良く見ると、全く同じように見えるのですが:
m08@makoto 09:22:43/100417(..work/scrc-2.1)% cat -n /usr/include/stdio.h| sed -n 221p    
   221  int      fgetpos(FILE * __restrict, fpos_t * __restrict);
m08@makoto 09:23:30/100417(..work/scrc-2.1)% cat -n ./src/bit/tst/compat.si | sed -n 140p   
   140  int fgetpos(FILE * __restrict, fpos_t * __restrict);
ここではないところを見ているのかな

#2 [NetBSD][Xen] NetBSD/amd64 5.0_STABLE + Xen 3.3.2

今これを入力している機械 (ML115 に NetBSD/amd64 5.0_STABLE) で Xen 3.3.2 を動かそうとしているが XEN3_DOM0 kernel を動かそうとすると、 多分その kernel を読終って、 初期化すると思われる時にすぐに再起動してしまう。 これは何かな、自分の設定の問題 ? 主記憶は 4GB。
menu=Boot normally:boot netbsd
menu=Boot single user:boot netbsd -s
menu=Disable ACPI:boot netbsd -2
menu=Disable ACPI and SMP:boot netbsd -12
menu=Drop to boot prompt:prompt
menu=Boot Xen with 1GB for dom0 (PC):load /netbsd-XEN3_DOM0
       multiboot /xen.gz dom0_mem=1024M
menu=Boot Xen with 1GB for dom0 (serial):load /netbsd-XEN3_DOM0.gz 
       console=com0 multiboot /xen.gz dom0_mem=1024M console=com1 com1=115200,8n1
default=1
timeout=10
これの上から 6 番目。 間違って いた
説明: load ; option 
のような構文になるところ、区切の ; が 抜けていた。今は、次のようにしている。 実際は一行。
menu=Boot Xen with 1GB for dom0 (PC): 
  load /netbsd-XEN3_DOM0 console=pc; 
  multiboot /xen-debug.gz dom0_mem=1024M
これで無事 DOM0 は動いた。 次は DOMU の centos 44 で、まだ動いていない。
modena@makoto 21:23:39/100416(~)% sudo xm list    
   (中略)
Name                 ID   Mem VCPUs      State   Time(s)
Domain-0              0  1024     1     r-----    252.6
hvm-centos44          7  1280     1     -b----    218.7

modena@makoto 21:23:54/100416(~)% sudo xm console 7 (中略) xenconsole: Could not read tty from store: No such file or directory

となるだけで、それ以上何も起きないのは ? .. 別の OS を試して見よう。

2010年04月15日(木) 旧暦 [n年日記] [更新:"2010/04/16 14:46:42"]

#1 [benchmark] Xeon って色々ある

DL360/G4 が 思ったほど速くない 件で /etc/fstab に次のような行を追加して /tmp を tmpfs にして見たら、
tmpfs                   /tmp    tmpfs   rw
(write cache はなくても) CPU はそれなりの稼動をしているが、やはり速くないと確認した。
dl360@makoto 13:28:18/100414(/export/src)% \
time sudo ./build.sh -j 8 release >& log-release-2
17501.40s 2787.94s 1:37:20.23 347% 0+0k 4879+26616io 37110033pf+0w
openssl speed rsa
                  sign    verify    sign/s verify/s
rsa  512 bits 0.000894s 0.000102s   1118.2   9850.7
rsa 1024 bits 0.004435s 0.000250s    225.5   3997.7
rsa 2048 bits 0.026247s 0.000775s     38.1   1290.3
rsa 4096 bits 0.171525s 0.002700s      5.8    370.4
dl360@makoto 22:58:24/100414(~)%
なお、tmpfs を /tmpfs に載せておいて、更に /export の上にその /tmpfs を mount_union で載せる件は、
mount_union: /tmpfs on /export: Operation not supported
で、だめだった。

T105 の場合:

build.sh が(うちでは) 最速の T105 (Quad Opteron) では
pts/0:makoto@t-105 22:02:17/100414(/export/src-5)> openssl speed rsa
Doing 512 bit private rsa's for 10s: 18551 512 bit private RSA's in 10.00s
Doing 512 bit public rsa's for 10s: 158288 512 bit public RSA's in 10.01s
Doing 1024 bit private rsa's for 10s: 4241 1024 bit private RSA's in 10.01s
Doing 1024 bit public rsa's for 10s: 73064 1024 bit public RSA's in 10.01s
Doing 2048 bit private rsa's for 10s: 767 2048 bit private RSA's in 10.01s
Doing 2048 bit public rsa's for 10s: 25369 2048 bit public RSA's in 10.01s
Doing 4096 bit private rsa's for 10s: 121 4096 bit private RSA's in 10.06s
Doing 4096 bit public rsa's for 10s: 7532 4096 bit public RSA's in 10.01s
OpenSSL 0.9.9-dev 09 May 2008
built on: NetBSD 5.0_BETA
options:bn(32,32) md2(int) rc4(4x,int) des(ptr,risc1,16,long) aes(partial) blowfish(idx) 
compiler: gcc version 4.1.3 20080704 (prerelease) (NetBSD nb2 20081120)
                  sign    verify    sign/s verify/s
rsa  512 bits 0.000539s 0.000063s   1855.1  15813.0
rsa 1024 bits 0.002360s 0.000137s    423.7   7299.1
rsa 2048 bits 0.013051s 0.000395s     76.6   2534.4
rsa 4096 bits 0.083140s 0.001329s     12.0    752.4
pts/0:makoto@t-105 22:15:41/100414(/export/src-5)> 
実は Xeon と言っても色々あって、以前の Netburst 系は遅い、と nork さんに 教わった。

CPU 当りでは最速の DL320/G5p:

pts/0:makoto@DL320 9:51:29/100415(~)> openssl speed rsa
Doing 512 bit private rsa's for 10s: 21985 512 bit private RSA's in 10.01s
Doing 512 bit public rsa's for 10s: 199185 512 bit public RSA's in 10.02s
Doing 1024 bit private rsa's for 10s: 4619 1024 bit private RSA's in 10.01s
Doing 1024 bit public rsa's for 10s: 79292 1024 bit public RSA's in 10.01s
Doing 2048 bit private rsa's for 10s: 794 2048 bit private RSA's in 10.01s
Doing 2048 bit public rsa's for 10s: 26358 2048 bit public RSA's in 10.01s
Doing 4096 bit private rsa's for 10s: 121 4096 bit private RSA's in 10.06s
Doing 4096 bit public rsa's for 10s: 7622 4096 bit public RSA's in 10.01s
OpenSSL 0.9.9-dev 09 May 2008
built on: NetBSD 5.0_RC3
options:bn(32,32) md2(int) rc4(4x,int) des(ptr,risc1,16,long) aes(partial) blowfish(idx) 
compiler: gcc version 4.1.3 20080704 (prerelease) (NetBSD nb2 20081120)
                  sign    verify    sign/s verify/s
rsa  512 bits 0.000455s 0.000050s   2196.3  19878.7
rsa 1024 bits 0.002167s 0.000126s    461.4   7921.3
rsa 2048 bits 0.012607s 0.000380s     79.3   2633.2
rsa 4096 bits 0.083140s 0.001313s     12.0    761.4
pts/0:makoto@DL320 9:52:56/100415(~)> 
こちらは、速い方の Xeon、多分 Core 系というものなのだろう

Comments related this article

Re: Xeon って色々ある by soda    2010/04/15 13:17
G5 は、PC で使われている Core Duo と同一のマイクロアーキテクチャで、速いです。G6...

#2 [Emacs] elscreen-1.4.6 on Emacs-23.1.90 (void-variable dir)

何故か、最近、次のように言われる
Debugger entered--Lisp error: (void-variable dir)
  elscreen-e21-command-line()
  command-line-1(("elscreen"))
  command-line()
  normal-top-level()
次のような回避策をして見た
--- elscreen.el~	2007-12-30 00:12:05.000000000 +0900
+++ elscreen.el	2010-04-15 08:47:18.000000000 +0900
@@ -1691,6 +1691,9 @@
                   ("-e"                . elscreen-command-line-funcall))))
 
   (static-when elscreen-on-emacs
+    (defvar dir command-line-default-directory)
+    (defvar line 0)
+    (defvar column 0)
     (defun elscreen-e21-command-line ()
       (when (string-match "\\`-" argi)
         (error "Unknown option `%s'" argi))
本来の直し方


2010年04月14日(水) 旧暦 [n年日記] [更新:"2010/04/15 08:00:01"]

#1 [benchmark] openssl speed rsa

Athlon x 2 4450B 2.4GHz
cpu1 at mainbus0 apid 1: AMD 686-class, 2400MHz, id 0x40f33 ML115
modena@makoto 18:51:49/100414(..gcc/work)% openssl speed rsa 
Doing 512 bit private rsa's for 10s: 30399 512 bit private RSA's in 8.83s
Doing 512 bit public rsa's for 10s: 206065 512 bit public RSA's in 6.87s
Doing 1024 bit private rsa's for 10s: 6261 1024 bit private RSA's in 6.29s
Doing 1024 bit public rsa's for 10s: 102444 1024 bit public RSA's in 6.53s
Doing 2048 bit private rsa's for 10s: 1219 2048 bit private RSA's in 6.66s
Doing 2048 bit public rsa's for 10s: 33601 2048 bit public RSA's in 5.64s
Doing 4096 bit private rsa's for 10s: 163 4096 bit private RSA's in 5.56s
Doing 4096 bit public rsa's for 10s: 11735 4096 bit public RSA's in 6.09s
OpenSSL 0.9.9-dev 09 May 2008
built on: NetBSD 5.0_STABLE
options:bn(64,64) md2(int) rc4(1x,char) des(idx,cisc,4,int) aes(partial) blowfish(idx) 
compiler: gcc version 4.1.3 20080704 (prerelease) (NetBSD nb2 20081120)
                  sign    verify    sign/s verify/s
rsa  512 bits 0.000290s 0.000033s   3442.7  29994.9
rsa 1024 bits 0.001005s 0.000064s    995.4  15688.2
rsa 2048 bits 0.005463s 0.000168s    183.0   5957.6
rsa 4096 bits 0.034110s 0.000519s     29.3   1926.9
modena@makoto 21:28:14/100414(..gcc/work)% 
こういう測り方もあるのか。以前にも教えてもらった気もするけれど。

Core i7な、MacBook pro で


2010年04月13日(火) 旧暦 [n年日記] [更新:"2010/04/16 00:19:38"]

#1 [機械] DL320/G3 -> ML110 -> DL360/G4 (Raid 考)

以前 (2010/02/03 の日記) から書いているが、最近は HP の DL320/G3 が結構気に入っている。
最近では五台 に増殖している。これらは一台に二台の SATA disk が挿せる。 それで、その disk のひき出しの数を数える、つまり 挿せる数と、挿す方のマウンタの数を較べると、 この他に DL320/G5p もあるので、それを含めると、二台足りない。 で、このマウンタだけ入手出来るかというと、そう簡単でもない (disk が入っている新品は入手可能だが 2 万円以上する)。

それで、たまたま ML110 に三つ入っているジャンク というのを見つけて、落札した (2,700 + (送料) 1,000) * 1.05 = 3,885 円)。 ところが、これがちっともジャンクでなく、 動いてしまった (期待外れ嬉というのかな)。
つまり、手元にあったマウンタも加えて 500GB x 4 を挿すと、RAID で動いているらしい。かなり筐体が大きいのが欠点であるが、 これはこれで便利そうだ。今の主ファイルサーバを rsync -aH で写して見た。
でもマウンタを二つ入手した筈が、更に一つ減ってしまった。(必要数が三になってしまった)

ML110 -> Adaptec 2610SA SATA-RAID:

それで、以上だと、二個入手するはずが、一つ失なってしまったという話である

で話は変って、その ML110 に内蔵されているのは Adaptec 2610SA SATA-RAID という基板である。これを使うようにするには、 電源投入時に鍵 (key) の組合せを入力すると設定画面が出る。

この設定をこまかく制御するには pkgsrc/sysutils/dptutil を使うらしい。 しかしこれが make 出来ない。 too old coding style (by obache) ということらしい。 直せば、とも思うが、この板は最大限 2TB ということで古いものらしく、 やる気が失せる(やる気があっても、僕に出来るものかは疑わしいけれど)

DL320/G3 の x2 disk の使い方は間違い ?:

この辺から気が付いた話があって、 今まで DL320/G3 の二つの disk は別々のもの (file system) として設定しているが、 多分それは期待された使い方ではなく、二つの disk を RAID にして使うのが普通なのでは、と思えて来た。

Software RAID or Hardware RAID:

RAID (例えば RAID5) は、個人的には気に入っている。4 台ある内の 1 台がこわれて も、それに気が付いた後で修復が可能だからである。 例えば TeraStation で RAID5 を使っている。ただ、Software RAID は、CPU で RAID をしている訳なので、 例えば書いている時に panic とかすると、 論理的にこわれてしまう。それでも修復はする。 ただ、遅い CPU だと、時間がかかる (上記 TeraStation だと当然容量にも依るが、20 時間とか、24 時間とか、かかる)
大昔 (1978年頃)に、CPU が M6800 clock 1MHz な Exorcisor という開発装置があって、 これの FDC (Floppy Disc Controller) が 6821 (Parallel i/f) 経由なソフトウェア制御だった。 間違って Reset ボタンを押したり、電源が不用意に切れると、 8 inch Floppy disk がこわれてしまう (初期化すれば、また使えるけれど)。
Software RAID だと、(修復可能な点は優位だけど)、まあ、以たようなものかと思う。
それに較べて Hardware RAID は、心強いのだけれど、しかし HW に依存すると、OSS (Open Source Software) 的には、 また別の努力 (driver + utility ) が必要となり (場合によっては情報不足で) 問題が発生した時に座礁してしまうこともあるのかなと思う。

DL360/G4 + レール:

DL320/G3 をラックに収納する時にはレール (HP part # 360332-003) があるとうれしい。今は(某所用に) 一組しか用意出来ていない。これは DL360 とも共通ということで、 つい DL360 G4 を見てしまったら Xeon 3.0GHz (dual core) x 2 というのが目に入り、 レールも付いているし、 build.sh を短時間で実行出来るかなということで、 SCSI Smart Array なのに手を出してしまった (送料共 7,040 円)。
ワットチェッカーで測ると 220W くらい喰っている。VA も同様。 負荷が大きい時には、300 Wくらいになることもある。 2 CPU なので、こんなものかなとも思うけれど、大喰い。 (実は DL360 の AC 入力は二つあって、ワットチェッカーが二つ要るのでは、 と某所に書いてしまったが、 二個口の AC 分岐を使れば良い、と某氏に教わった ... 何とも自分の愚かさにあきれ ...)
さて、問題は disk cache を有効にするには
Also available as an optional upgrade is the
128MB Battery Backed Write Cache Enabler Option Kit (PN 346914-B21)
が必要らしい。これがないと build.sh に 2 時間以上かかってしまう。
dl360@makoto 16:04:41/100404(/export/src)%
time sudo ./build.sh -j 4 release> ../log-release
10533.01s 1754.46s 2:26:15.94 140% 0+0k 2483+726040io 47348pf+0w
別の考え方として、 今は主記憶が 2GB なので、これを 4GB くらいに増せば主記憶上で速く実行出来るかも知れない。(そのままでも行ける ?) 現在は
1GB DDR PC2700 CL2.5 ECC
P/N 331562-051
が二個入っている
tmpfs を /tmpfs に mount しておいて、それを更に mount_union 出来るのかな。

全く別件 ( 1TB disk 回収):

1/22/2010 にヨドバシカメラで 8,980 円で 買った 1TB の WD10WEARS が、都内某所で動作中におかしくなったのを修理に出してあったが、 同品交換で返って来たのを受取る。


2010年04月12日(月) 旧暦 [n年日記] [更新:"2010/04/19 00:23:55"]

#1 [無線] JIDX CW 2010 (SOSB/40m)

JIDX CW 2010 に参加した、部門は Single Operator Single Band 40m Low Power ( SOSB/40m LP) である 今回は深夜に Europe から呼ばれて、とても楽しかった。 寝たのは(一応) 日曜の 05:00-10:00 だけかな。 350 局 59 Entity 29 zone くらい。
N1MM によれば、運用時間は 16:03 と出ている。 時間別の交信数は次のようなもの


#2 [無題] 芝生を植えて見た

試しに、高麗芝を一平米分、400 円足らずで買って来て並べて見た。 10 枚である。



最近の日記
2024年04月29日
dkim
2024年03月10日
停電 (瞬電)
2024年03月03日
the second try on bare-metal
useradd
2024年02月29日
opendkim and senmail
2024年01月24日
chat/iam 0.0.8
以上、9 日分です。
タイトル一覧
カテゴリ分類
Powered by hns-2.19.9, HyperNikkiSystem Project

Count.cgi (since 2000/02/05)