Google Groups no longer supports new Usenet posts or subscriptions. Historical content remains viewable.

Dismiss

stable-digest V5 #783

3 views

Skip to first unread message

owner-freebsd...@freebsd.org

unread,

Feb 3, 2003, 4:59:46 PM2/3/03

stable-digest Monday, February 3 2003 Volume 05 : Number 783

In this issue:
疹蜩蝈镱塍麒螯 WebMoney 镱黩�徨耧豚蝽�
Re: at33 vs others?
Re: at33 vs others?
ffs_update panic with latest kernel ... softupdates problem?
超一流网络视频会议、免费电话、远程教学专家会诊系统
SiS-Patch inside CVS
Cirrus - Which driver ?
No terminfo in system ncurses
Re: Cirrus - Which driver ?
Re: No terminfo in system ncurses
libc_r seems to be broken in 4.7-STABLE as of Jan 27
RE: ffs_update panic with latest kernel ... softupdates problem?
ata problems
Re: Via Rhine owners, please test!
HEADS UP: OpenSSH 3.5p1
Re: Via Rhine owners, please test!
recommendations on the newfs of a 1.0TB fs...
Re: recommendations on the newfs of a 1.0TB fs...
Re: recommendations on the newfs of a 1.0TB fs...
Re: recommendations on the newfs of a 1.0TB fs...
Re: recommendations on the newfs of a 1.0TB fs...
Re: recommendations on the newfs of a 1.0TB fs...
Re: recommendations on the newfs of a 1.0TB fs...
"ping: sendto: No buffer space available"
Re: recommendations on the newfs of a 1.0TB fs...
Re: ata problems

----------------------------------------------------------------------

Date: Sun, 2 Feb 2003 23:07:32 +0300
From: 人苓 <ilya...@mail.ru>
Subject: 疹蜩蝈镱塍麒螯 WebMoney 镱黩�徨耧豚蝽�

�珥帼, 赅�钺爨眢螯耔耱屐�WebMoney. �礤镳邃豚汔�忄�镳钽疣祆� 觐蝾瘥�镥疱忸��溴睃汨磬忄�觐朦尻.
入�耧桉觇觐朦觐� 觐蝾瘥�忸玮疣 忄 溴睃汨 �礤耜铍�疣�犷朦. �镳邃豚汔�锣�溴轳蜮栩咫�疣犷蜞栝耧耦�漕徉怆屙� 溴礤�磬忄�觐朦尻.
篷腓忄�桧蝈疱耥� 蝾沅�镨蝈 mailto:webmone...@ukr.net?subject=WebMoney 镱黩�徨耧豚蝽�
------------------------------

Date: Sun, 2 Feb 2003 17:24:37 -0500
From: Bill Vermillion <b...@wjv.com>
Subject: Re: at33 vs others?

> Date: Sun, 2 Feb 2003 11:10:59 +0100 (CET)
> From: Alban Hertroys <dal...@solfertje.student.utwente.nl>
> Subject: Re: ata33 vs others?
>
> On 31 Jan, Eric Timme punched keys in this particular order:
> > After wrestling with an 80gb hd and an Abit BX6-R2 for an afternoon and being
> > pleasantly surprised that Abit's last bios release would allow my board to
> > detect and use an 80gb hd I came to the sad realization that the computer
> > only supported ata33.
>
> That shouldn't matter much if it is the only drive on the IDE channel.
>
> Modern harddrives still perform under 40MB/s last time I checked (the
> IBM 60GXP series could do 37.5MB/s), and that is at maximum.
>
> The ATA100 standard was necessery because two disks on a channel could
> in some cases press the required data throughput of the controller
> over 66MB/s (2x37.5=75), though I'm sure it maxes out quite a bit under
> 100MB/s.
>
> I don't know how much overhead data on an IDE channel has (like 'packet
> headers' or some equivalent) and we now have an ATA133 standard that
> would be kind of absurd with the above reasoning.
> Maybe someone can shine a light on this?
>

I just updated an iNTEL 1100R rack mount from an ATA33 [don't
recall the drive] to a Promise 100/133 controller and a Maxtor
120GB 7200 RPM ATA133 I saw significant increase.

I used the old iozone2 [in ports] as I've used it for years
and it measure r/w performance through the file system.

The old system, using a 200MB fiel for test showed
9,166,619 B/sec write and 9,422,339B/sec read.

With the new combo on a fully fresh filesystem I used
a 500MB file for test.

I got 21,408,22B/sec write and 46,091,252B/sec write.

Since this was so much faster I repeated the test with
a 1GB file. That gave me this:

20,677,512B/sec write and 45,554,310Bsec read.

I just compiiled 4.7-P3 [CPU is 650MHz] and came in about 55
minutes and the kernel compile took just about 9 minutes.

The faster drive and controller make a significant perfomance
difference.

Bill

------------------------------

Date: Mon, 03 Feb 2003 10:09:25 +1100
From: David Burren <da...@burren.cx>
Subject: Re: at33 vs others?

> I just updated an iNTEL 1100R rack mount from an ATA33 [don't
> recall the drive] to a Promise 100/133 controller and a Maxtor
> 120GB 7200 RPM ATA133 I saw significant increase.

Yes ATA133 can be faster as it removes one bottleneck, but the drive
will be making a big difference too. The 7200 RPM of the Maxtor
will be boosting the transfer rate, and the larger buffer, (possibly)
different buffer management in the firmware, and (presumably) lower
seek times of the disk will all be contributing. The position of the
filesystem partition on the drive and the fresh filesystem will also
affect the result.

Your numbers are not a true indication of the advantages of ATA133.
Try connecting the drive with an ATA33 cable to drop the bus speed
back and do your test again. That would be a better comparison.

But then you might not care that much. Your system is significantly
faster than it was, and that's probably the bottom line for most
people. :-)
__
David B

------------------------------

Date: Sun, 2 Feb 2003 22:32:39 -0400 (AST)
From: "Marc G. Fournier" <scr...@hub.org>
Subject: ffs_update panic with latest kernel ... softupdates problem?

Just upgraded my world/kernel today, and it panic'd within 12hrs with:

(kgdb) where
#0 0xc7c1a256 in ?? ()
#1 0xc014e46c in boot (howto=256) at /usr/src/sys/kern/kern_shutdown.c:223
#2 0xc014eb91 in panic (fmt=0xc022fc99 "%s") at /usr/src/sys/kern/kern_shutdown.c:595
#3 0xc01ffb29 in trap_fatal (frame=0xea667e94, eva=66) at /usr/src/sys/i386/i386/trap.c:974
#4 0xc01ff795 in trap_pfault (frame=0xea667e94, usermode=0, eva=66) at /usr/src/sys/i386/i386/trap.c:867
#5 0xc01ff2ef in trap (frame={tf_fs = -362414056, tf_es = -1072234480, tf_ds = -737476592, tf_edi = 0, tf_esi = -879852544, tf_ebp = -362381588, tf_isp = -362381632,
tf_ebx = -946393088, tf_edx = -1, tf_ecx = -296658752, tf_eax = 0, tf_trapno = 12, tf_err = 0, tf_eip = -1071952598, tf_cs = 8, tf_eflags = 66118, tf_esp = 0,
tf_ss = -296658752}) at /usr/src/sys/i386/i386/trap.c:466
#6 0xc01b4d2a in ffs_update (vp=0xee5158c0, waitfor=0) at /usr/src/sys/ufs/ffs/ffs_inode.c:105
#7 0xc01bd267 in ffs_sync (mp=0xc7958600, waitfor=3, cred=0xc62fba00, p=0xe7c9a780) at /usr/src/sys/ufs/ffs/ffs_vfsops.c:1025
#8 0xc017e8dc in sync_fsync (ap=0xea667f7c) at /usr/src/sys/kern/vfs_subr.c:2972
#9 0xc017cbb3 in sched_sync () at vnode_if.h:558
(kgdb) up 5
#5 0xc01ff2ef in trap (frame={tf_fs = -362414056, tf_es = -1072234480, tf_ds = -737476592, tf_edi = 0, tf_esi = -879852544, tf_ebp = -362381588, tf_isp = -362381632,
tf_ebx = -946393088, tf_edx = -1, tf_ecx = -296658752, tf_eax = 0, tf_trapno = 12, tf_err = 0, tf_eip = -1071952598, tf_cs = 8, tf_eflags = 66118, tf_esp = 0,
tf_ss = -296658752}) at /usr/src/sys/i386/i386/trap.c:466
466 (void) trap_pfault(&frame, FALSE, eva);
(kgdb) frame frame->tf_ebp frame->tf_eip #0 0xc01b4d2a in ffs_update (vp=0xee5158c0, waitfor=0) at /usr/src/sys/ufs/ffs/ffs_inode.c:105
105 if (DOINGSOFTDEP(vp))
(kgdb) list
100 (int)fs->fs_bsize, NOCRED, &bp);
101 if (error) {
102 brelse(bp);
103 return (error);
104 }
105 if (DOINGSOFTDEP(vp))
106 softdep_update_inodeblock(ip, bp, waitfor);
107 else if (ip->i_effnlink != ip->i_nlink)
108 panic("ffs_update: bad link cnt");
109 *((struct dinode *)bp->b_data +

------------------------------

Date: Mon, 3 Feb 2003 13:38:01 +0800
From: 匿名 <>
Subject: 超一流网络视频会议、免费电话、远程教学专家会诊系统

超一流网络视频会议、免费电话、远程教学专家会诊系统
本公司最新产品:网络视频会议视频服务器---广泛用于高清晰度局域或广域视频会议、网络监控、幼儿园敬老院和医院病房家人异地远程监护、专家远程医疗会诊和教学、网络远程教室以及无话费可视化客户服务系统,网上风景区,酒店实时视频展示、电视节目网上实时转播、大型展会现场网络直播等,网上传输效果居目前国内外领先水平，价格仅有进口同档次产品1/10。
欢迎网上观看演示效果:www.webht.com

演示环境及试用方法：
本公司视频会议服务器每天24小时连续开机，若有兴趣请在本公司网站下载最新版绿色客户端软件,直接点击执行后输入服务器号3300002，用户名和密码均为“1234”，按确定后稍等片刻即可连通视频会议系统看到演示现场图象，在此处我们可用话筒进行视频会议语音交谈，亲身体验最新IT成果，欢迎与任何国外产品作一个比较，您会发现中国的月亮比外国的月亮还要圆。

应用之一 ------ 网络视频远程教学和专家远程会诊系统：
名校名师讲课、高级教授讲学、医学手术权威的手术都可使用本系统通过网络向特定用户开放，获有密码的教室或个人可登陆系统参加学习。一套服务器可供10---100以上用户同时登陆，只要服务器端带宽足够（500K X 用户数量），用户端300-500K带宽无路由延迟，可做到视音频基本同步，双工对讲。

专家远程会诊———利用视讯业务实现中心医院与基层医院就疑难病症进行会诊、指导治疗与护理、对基层医务人员的医学培训等等。高质量的视讯业务使医生、护士在不同地方同时协同工作成为可能。远程医疗对于一些中小医院有着重要的意义，可以得到大医院的医学专家的咨询和会诊。

应用之一 ------ 无话费可视化客户服务系统：
总公司只需安装一台视频服务器、几只摄像机，远程的各地连锁店分公司只需使用免费的客户端软件通过电脑上网就可通过网络与总部对话，同时看到总部高清晰图象。可用于总部对下属单位召开视频会议、下属呼叫总部进行业务联系或总部新产品对国内外远端客户进行演示、培训和电子白板文字沟通，这一特点就使得本系统可取代大型企业集团目前昂贵的800电话客户服务系统而通过廉价的网络方式开展用户服务,使用户直接看到产品实物达到最好的宣传服务作用，并且双方都不会产生任何话费。

其它产品：
1、视频双绞线收发器--廉价的高性能远距离监控传输设备,距离可达1-3千米。
2、LED电子显示屏，监控报警专用显示条屏；3、全中文矩阵万能切换监控报警主机 4、网络视频会议系统 5、销售保安监控、楼宇对讲设备，承接保安监控和校园网络工程。7、IC卡出租车计价器和公务汽车管理机；8、彩色、黑白摄象机，水下摄象机,系列红外灯、汽车夜视仪、系列无线影音收发器、光端机、视频共缆传输等设备.

本系统特点：
在同一个网络带宽环境下图象和语音质量可超过目前所有的国内外产品，本系统实现高速高效率的软件压缩技术，使实时压缩传输的图象为756X576象素，而非象素的放大，故图象清晰度高，使用鼠标点击画面可复原放大为满屏显示，使用大屏幕彩电或投影显示满屏图象会有意想不到的高清晰度效果。

最新消息：专为本系统配用的MCU业已开发试验成功，已通过16、32和64和128 用户的大系统测试，转发的图象清晰稳定，多台MCU级连后可构成更大系统，详细组网方按请函电联系。

徐州华统电脑设备有限公司徐州市青年路288号四楼 TEL：0516-3738506 2682031
FAX：0516-3752314 联系人：王瑞海网址：www.webht.com 邮箱：xzh...@163.com

欢迎光临网上营销软件超市 http://www.sesoft.findhere.org
本邮件用超级广告软件发送 http://www.sesoft.findhere.org

------------------------------

Date: Mon, 3 Feb 2003 10:00:17 +0100 (CET)
From: "Jan Siml" <j...@jules.de>
Subject: SiS-Patch inside CVS

This is a multi-part message in MIME format.

- --bound1044262817
Content-Type: text/plain; charset=iso-8859-1
Content-Transfer-Encoding: 7bit

Hello Luogi!

I doubt, if your patch* is already integrated into the code, that is delivered
via CVS. The problem is the following. I patched a fresh installed FreeBSD
4.7-RELEASE with your patchfile. The onbord nic works fine. But what if I fetch
sources via CVS? Could i make an build-/installkernel without fear that the nic
(SiS961/961B/962) won't work after this step?

* http://people.freebsd.org/~luoqi/sis.diff

Regards

Jan

P.S. Please CC me, as I'm not subscribed to the freebsd-stable mailing list.

- --bound1044262817--

------------------------------

Date: Mon, 3 Feb 2003 10:26:01 +0100
From: Pierrick Brossin <pbro...@swissgeeks.com>
Subject: Cirrus - Which driver ?

Hi,

I have a CL-GD5446 (Cirrus) Graphic Card and was wondering which driver I should
use!

First I tried the cirrus driver but the screen is not OK. I use WindowMaker and
it appears two times on the screen. I've tried to modify VertSync and
HorizRefresh with no success. I also tried to insert a Modeline.

I surfed the net and found that XFree recommends to use the SVGA driver for
Cirrus Cards.. but svga_drv.o is not in the list of drivers in
/usr/X11R6/lib/modules/drivers/

Any idea what I should do ?

Thanx

- --
Pierrick Brossin
IT Swiss - QUARK Media House
6a Puits Godet, 2000 Neuchatel, Switzerland
Tel Prof: +41-327201423 Mobile Priv: +41-794137145
Mail Prof: pbro...@quark.ch Mail Priv: ad...@swissgeeks.com

------------------------------

Date: Mon, 3 Feb 2003 07:39:51 -0500
From: AlanE <al...@geeksrus.net>
Subject: No terminfo in system ncurses

Why is there no full terminfo db in the system ncurses? You have to install
the port in order to get it. Seems wrong to me.

- --
AlanE (Alan Eldridge)
Unix/C(++) IT Pro for 20 yrs, likes fixing weird distributed systems bugs.
KDE, KDE-FreeBSD Teams (http://www.kde.org, http://freebsd.kde.org/)

------------------------------

Date: Mon, 3 Feb 2003 14:26:16 +0100
From: Pierrick Brossin <pbro...@swissgeeks.com>
Subject: Re: Cirrus - Which driver ?

Quoting Dave Uhring <duh...@charter.net>:

> Executing, as root, XFree86 -configure should provide you with an
> XF86Config.new file with the proper driver declared.

Nop it creates a file with VGA driver :(

------------------------------

Date: Mon, 3 Feb 2003 16:20:37 +0200
From: Vallo Kallaste <ka...@estpak.ee>
Subject: Re: No terminfo in system ncurses

On Mon, Feb 03, 2003 at 07:39:51AM -0500, AlanE <al...@geeksrus.net> wrote:

> Why is there no full terminfo db in the system ncurses? You have to install
> the port in order to get it. Seems wrong to me.

It's stripped out for historical reasons I guess. BSD systems have
used to use termcap traditionally, but you probably know it already.
So the question is why the history forces us to termcap, so I have
to install /usr/share/misc/terminfo by hand.. Good ground for new
(and at the same time old) holy war, isn't it :-/
- --

Vallo Kallaste
ka...@estpak.ee

------------------------------

Date: Mon, 03 Feb 2003 16:19:50 +0100
From: =?ISO-8859-1?Q?Jonas_B=FClow?= <jo...@servicefactory.se>
Subject: libc_r seems to be broken in 4.7-STABLE as of Jan 27

Hi,

There seems to be a problem in libc_r causing processes to hang
(spinning) during startup of the process.

It happens when the box is very busy. I got the back trace below using
gcore on such a spinning process. The process consumed like 70% CPU.

Has this been seen before? What more can I do to help solve the problem?

The is a diskless box without swap. It's likely it went out of memory.

(gdb) bt
#0 0x280c5660 in _thread_sig_handle_pending () from /usr/lib/libc_r.so.4
#1 0x280c50eb in _thread_sig_handler () from /usr/lib/libc_r.so.4
#2 0xbfbfffac in ?? ()
#3 0x281000f8 in _thread_leave_cancellation_point () from
/usr/lib/libc_r.so.4
#4 0x28100445 in _thread_leave_cancellation_point () from
/usr/lib/libc_r.so.4
#5 0x28100b5d in malloc () from /usr/lib/libc_r.so.4
#6 0x280cbbd4 in _thread_init () from /usr/lib/libc_r.so.4
#7 0x2810157c in _thread_init_invoker::_thread_init_invoker () from
/usr/lib/libc_r.so.4
#8 0x281015b5 in _thread_init_invoker::_thread_init_invoker () from
/usr/lib/libc_r.so.4
#9 0x281015de in _thread_init_invoker::_thread_init_invoker () from
/usr/lib/libc_r.so.4
#10 0x2810160c in _thread_init_invoker::_thread_init_invoker () from
/usr/lib/libc_r.so.4
#11 0x28088caa in _init () from /usr/lib/libc_r.so.4
#12 0x2804f978 in _rtld () from /usr/libexec/ld-elf.so.1

/jonas

------------------------------

Date: Mon, 3 Feb 2003 12:08:00 -0500
From: "Allan Jude" <937...@primus.ca>
Subject: RE: ffs_update panic with latest kernel ... softupdates problem?

I also got a fss_update panic after updating my kernel/world

- -----Original Message-----
From: owner-free...@FreeBSD.ORG
[mailto:owner-free...@FreeBSD.ORG] On Behalf Of Marc G. Fournier
Sent: Sunday, February 02, 2003 9:33 PM
To: freebsd...@freebsd.org
Subject: ffs_update panic with latest kernel ... softupdates problem?

Just upgraded my world/kernel today, and it panic'd within 12hrs with:

(kgdb) where
#0 0xc7c1a256 in ?? ()
#1 0xc014e46c in boot (howto=256) at
/usr/src/sys/kern/kern_shutdown.c:223
#2 0xc014eb91 in panic (fmt=0xc022fc99 "%s") at
/usr/src/sys/kern/kern_shutdown.c:595
#3 0xc01ffb29 in trap_fatal (frame=0xea667e94, eva=66) at
/usr/src/sys/i386/i386/trap.c:974
#4 0xc01ff795 in trap_pfault (frame=0xea667e94, usermode=0, eva=66) at
/usr/src/sys/i386/i386/trap.c:867
#5 0xc01ff2ef in trap (frame={tf_fs = -362414056, tf_es = -1072234480,
tf_ds = -737476592, tf_edi = 0, tf_esi = -879852544, tf_ebp =
- -362381588, tf_isp = -362381632,
tf_ebx = -946393088, tf_edx = -1, tf_ecx = -296658752, tf_eax = 0,
tf_trapno = 12, tf_err = 0, tf_eip = -1071952598, tf_cs = 8, tf_eflags =
66118, tf_esp = 0,
tf_ss = -296658752}) at /usr/src/sys/i386/i386/trap.c:466
#6 0xc01b4d2a in ffs_update (vp=0xee5158c0, waitfor=0) at
/usr/src/sys/ufs/ffs/ffs_inode.c:105
#7 0xc01bd267 in ffs_sync (mp=0xc7958600, waitfor=3, cred=0xc62fba00,
p=0xe7c9a780) at /usr/src/sys/ufs/ffs/ffs_vfsops.c:1025
#8 0xc017e8dc in sync_fsync (ap=0xea667f7c) at
/usr/src/sys/kern/vfs_subr.c:2972
#9 0xc017cbb3 in sched_sync () at vnode_if.h:558
(kgdb) up 5
#5 0xc01ff2ef in trap (frame={tf_fs = -362414056, tf_es = -1072234480,
tf_ds = -737476592, tf_edi = 0, tf_esi = -879852544, tf_ebp =
- -362381588, tf_isp = -362381632,
tf_ebx = -946393088, tf_edx = -1, tf_ecx = -296658752, tf_eax = 0,
tf_trapno = 12, tf_err = 0, tf_eip = -1071952598, tf_cs = 8, tf_eflags =
66118, tf_esp = 0,
tf_ss = -296658752}) at /usr/src/sys/i386/i386/trap.c:466
466 (void) trap_pfault(&frame, FALSE, eva);
(kgdb) frame frame->tf_ebp frame->tf_eip
#0 0xc01b4d2a in ffs_update (vp=0xee5158c0, waitfor=0) at
/usr/src/sys/ufs/ffs/ffs_inode.c:105
105 if (DOINGSOFTDEP(vp))
(kgdb) list
100 (int)fs->fs_bsize, NOCRED, &bp);
101 if (error) {
102 brelse(bp);
103 return (error);
104 }
105 if (DOINGSOFTDEP(vp))
106 softdep_update_inodeblock(ip, bp, waitfor);
107 else if (ip->i_effnlink != ip->i_nlink)
108 panic("ffs_update: bad link cnt");
109 *((struct dinode *)bp->b_data +

------------------------------

Date: Mon, 3 Feb 2003 12:48:53 -0500
From: Peter Radcliffe <p...@pir.net>
Subject: ata problems

Updating to -STABLE as of yesterday morning the cdrom combo drive in
my laptop is not longer recognised.

ata1-master: ATA identify retries exceeded
ad0: 38154MB <IC25N040ATCS04-0> [77520/16/63] at ata0-master UDMA100
Mounting root from ufs:/dev/ad0s2a

If I use atacontrol to detatch and attach ata1 it appears;

acd0: CD-RW <HL-DT-STCD-RW/DVD DRIVE GCC-4160N> at ata1-master UDMA33
cd0 at ata1 bus 0 target 0 lun 0
cd0: <HL-DT-ST RW/DVD GCC-4160N 0012> Removable CD-ROM SCSI-0 device
cd0: 33.000MB/s transfers
cd0: Attempt to query device size failed: NOT READY, Medium not present

(I'm using atapicam).

- --
pir pir...@pir.net pir...@net.tufts.edu

------------------------------

Date: Tue, 4 Feb 2003 05:57:40 +1100
From: Rudolph Pereira <meme...@yahoo.com.au>
Subject: Re: Via Rhine owners, please test!

On Fri, Jan 31, 2003 at 10:59:28PM -0600, Mike Silbersack wrote:
>
> At the request of Thomas Nystrom, I have now committed his patches to fix
> the hanging problems of some rhine chipsets and his fixes to allow 6105
> (Rhine III) chips to work properly to -current. Naturally, I'm going to
> wait at least a week before MFCing this change to 4-stable. In the
> meantime, I'd like to request those who have had problems with Via Rhine
> cards (either hangs, or unsupported 6105 cards) to try out this patch and
> tell me how it works out. This patch contains all of the changes
> described above, and should apply cleanly to any relatively recent
> 4.7-stable machine.
This works fine for me. I have, according to pciconf, a "VT6102 Rhine
II" card, though it is an integrated one, and different enough to normal
6102 cards that it wasn't working before.

Thanks for your great work Thomas/Mike.

------------------------------

Date: Mon, 03 Feb 2003 20:22:33 +0100
From: Dag-Erling Smorgrav <d...@ofug.org>
Subject: HEADS UP: OpenSSH 3.5p1

I've upgraded OpenSSH in -STABLE to 3.5p1. Please report any breakage
directly to me.

DES
- --
Dag-Erling Smorgrav - d...@ofug.org

------------------------------

Date: Mon, 03 Feb 2003 20:39:07 +0100
From: dirk....@dinoex.sub.org (Dirk Meyer)
Subject: Re: Via Rhine owners, please test!

so far my problems seem to vanished in the last 2 days:
Feb 1 18:38:28 ceres /kernel: vr0: watchdog timeout
Feb 1 18:38:28 ceres /kernel: vr0: reset never completed!
Feb 1 18:38:28 ceres /kernel: vr0: reset never completed!

Feb 1 21:54:43 ceres /kernel: vr0: <VIA VT6102 Rhine II 10/100BaseTX> port 0xb800-0xb8ff mem 0xdd800000-0xdd8000ff irq 12 at device 10.0 on pci0
Feb 1 21:54:43 ceres /kernel: vr0: Ethernet address: 00:50:ba:eb:xx:xx
Feb 1 21:54:43 ceres /kernel: miibus0: <MII bus> on vr0

vr0@pci0:10:0: class=0x020000 card=0x14011186 chip=0x30651106 rev=0x42 hdr=0x00

kind regards Dirk

- - Dirk Meyer, Im Grund 4, 34317 Habichtswald, Germany
- - [dirk....@dinoex.sub.org],[dirk....@guug.de],[din...@FreeBSD.org]

------------------------------

Date: Mon, 3 Feb 2003 11:48:28 -0800
From: Ryan Dooley <ry...@third-man.com>
Subject: recommendations on the newfs of a 1.0TB fs...

Hey All,

I'm looking for recommendations on the newfs options for a 1.0TB file system
for say /home (it's a fileserver for 52000+ accounts that range from having
4k to 40GB in their directories.)

When I initally created the file system, the options I set are: block size of
65536 and a fragsize of 8192.

This has worked out for me in the past but now that the fs is more than 50%
in use, the fsck's (it's a -STABLE system) are taking about an hour to complete.

In a previous thread somebody recommended that I tweak the inodes when I have
the time to rebuild the filesystem (I might have that opportunity soon).

Currently, about 20% of the inodes (6141123/25124155 - used/free) are in use.

Any recommendations?

Thanks in advance,
Cheers,
Ryan

------------------------------

Date: Mon, 03 Feb 2003 21:05:33 +0100
From: Dag-Erling Smorgrav <d...@ofug.org>
Subject: Re: recommendations on the newfs of a 1.0TB fs...

Ryan Dooley <ry...@third-man.com> writes:
> When I initally created the file system, the options I set are: block size of
> 65536 and a fragsize of 8192.

How many cylinder groups?

> This has worked out for me in the past but now that the fs is more than 50%
> in use, the fsck's (it's a -STABLE system) are taking about an hour to complete.

The question is, why does it need fsck'ing in the first place?

DES
- --
Dag-Erling Smorgrav - d...@ofug.org

------------------------------

Date: Mon, 03 Feb 2003 15:28:04 -0500
From: Bill Moran <wmo...@potentialtech.com>
Subject: Re: recommendations on the newfs of a 1.0TB fs...

Ryan Dooley wrote:
> Hey All,
>
> I'm looking for recommendations on the newfs options for a 1.0TB file system
> for say /home (it's a fileserver for 52000+ accounts that range from having
> 4k to 40GB in their directories.)
>
> When I initally created the file system, the options I set are: block size of
> 65536 and a fragsize of 8192.
>
> This has worked out for me in the past but now that the fs is more than 50%
> in use, the fsck's (it's a -STABLE system) are taking about an hour to complete.

I'd consider fixing the situation thats causing it to need fscked before doing
anything else!

> In a previous thread somebody recommended that I tweak the inodes when I have
> the time to rebuild the filesystem (I might have that opportunity soon).
>
> Currently, about 20% of the inodes (6141123/25124155 - used/free) are in use.

Sounds like you have enough users to have a pretty good idea of the average
bytes per file. Use -i to specify that when you newfs. Don't overdo it! It's
pretty frustrating to have 400G left on the drive an no inodes left to create
any new files, but if you've got 50% of the space full and only 20% of the inodes,
you can definately adjust this some.

There's also -g and -h options. I'm not sure what effect these have, however.

- --
Bill Moran
Potential Technologies
http://www.potentialtech.com

------------------------------

Date: Mon, 3 Feb 2003 12:45:24 -0800
From: Ryan Dooley <ry...@third-man.com>
Subject: Re: recommendations on the newfs of a 1.0TB fs...

> I'd consider fixing the situation thats causing it to need fscked before
> doing
> anything else!

Heh.. yeah. Normally my box doesn't crash. The last time the box was
down for any thing other than a maintaince window was over a year ago. I'd
say the box was rock solid :-)

> Sounds like you have enough users to have a pretty good idea of the average
> bytes per file. Use -i to specify that when you newfs. Don't overdo it!
> It's
> pretty frustrating to have 400G left on the drive an no inodes left to
> create
> any new files, but if you've got 50% of the space full and only 20% of the
> inodes,
> you can definately adjust this some.

I'm wondering what values I might try for -i that might be reasonable.

> There's also -g and -h options. I'm not sure what effect these have,
> however.

Yeah, I saw those options as well, but since this is an academic environment,
"average" doesn't apply :-)

Cheers,
Ryan

------------------------------

Date: Mon, 3 Feb 2003 12:46:44 -0800
From: Ryan Dooley <ry...@third-man.com>
Subject: Re: recommendations on the newfs of a 1.0TB fs...

> How many cylinder groups?

Probably would help if I replied to the list as well.

1421 cpg.

Cheers,
Ryan

------------------------------

Date: Tue, 4 Feb 2003 00:07:52 +0300 (MSK)
From: Andrey Alekseyev <ui...@zenon.net>
Subject: Re: recommendations on the newfs of a 1.0TB fs...

> I'm wondering what values I might try for -i that might be reasonable.

I believe, you should examine your existing file system more closely.

Recently we had a task of optimizing a large filesystem (300+ GB) for our
large mail cluster (hundreds of thousands of mail accounts and about 10000+ mail
domains). The filesystem itself was supposed to contain about 1500000
small files and about 400000 directories.

For that task I had written a rather simple utility that traverses a
directory tree and produces a statistics report on the number of files for
each given range of size (less than 8KB, 8KB < n < 16KB, greater than
16KB, etc.). Also, the total number of files and directories is accumulated
along with other interesting detail. I must say it was rather a useful
information and food for thought before doing any optimization! :)

However, I should mention, the OS is Solaris 8 and the filesystem is not UFS.
I was optimizing given certain details on file system allocation mechanisms.
Especially, block size was the question.

- --
Andrey Alekseyev. Zenon N.S.P.
Senior Unix system administrator

------------------------------

Date: Mon, 03 Feb 2003 16:13:23 -0500
From: Bill Moran <wmo...@potentialtech.com>
Subject: Re: recommendations on the newfs of a 1.0TB fs...

Ryan Dooley wrote:
>>I'd consider fixing the situation thats causing it to need fscked before
>>doing
>>anything else!
>
> Heh.. yeah. Normally my box doesn't crash. The last time the box was
> down for any thing other than a maintaince window was over a year ago. I'd
> say the box was rock solid :-)

Well that's good to know.

>>Sounds like you have enough users to have a pretty good idea of the average
>>bytes per file. Use -i to specify that when you newfs. Don't overdo it!
>>It's
>>pretty frustrating to have 400G left on the drive an no inodes left to
>>create
>>any new files, but if you've got 50% of the space full and only 20% of the
>>inodes,
>>you can definately adjust this some.
>
> I'm wondering what values I might try for -i that might be reasonable.

Do the math. df -hi will tell you the number of bytes and inodes used.
Do a simple bytes/inodes and add about 10% just to be sure.
That should take care of you.

There are no "reasonable" values in my opinion. I have one client that
makes a lot of files for robot machines (basically, text files). They
don't get much longer than ~15K. If I redid their filesystem, I'd use
4000 bytes/inode. I have another client that's a graphic design firm,
their server is filled with photo-quality tiffs and jpegs. I used
500000bytes/inode and the free space is still disappearing faster than
the inodes are. If your server is 50% full, you should have enough data
to make a reasonable estimate. Just pad it a little so you don't run
out of inodes before you do free space.

- --
Bill Moran
Potential Technologies
http://www.potentialtech.com

------------------------------

Date: Mon, 3 Feb 2003 16:23:49 -0500
From: Peter Radcliffe <p...@pir.net>
Subject: "ping: sendto: No buffer space available"

ping: sendto: No buffer space available
ping: sendto: No buffer space available

I see this occasionally on my -stable box and havn't had a good
explanation of why and what buffer. Searching isn't turning up
anything useful.

Is there something I can tweak to make this less likely ?

- --
pir pir...@pir.net pir...@net.tufts.edu

------------------------------

Date: Mon, 03 Feb 2003 22:34:40 +0100
From: Dag-Erling Smorgrav <d...@ofug.org>
Subject: Re: recommendations on the newfs of a 1.0TB fs...

Bill Moran <wmo...@potentialtech.com> writes:
> Do the math. df -hi will tell you the number of bytes and inodes used.
> Do a simple bytes/inodes and add about 10% just to be sure.
> That should take care of you.

You should *subtract* to be sure. Specify to newfs an average file
size which is slightly lower than the measured average file size, so
it creates slightly more inodes than you need.

DES
- --
Dag-Erling Smorgrav - d...@ofug.org

------------------------------

Date: Mon, 3 Feb 2003 19:35:30 +0100
From: Steve O'Hara-Smith <st...@sohara.org>
Subject: Re: ata problems

On Mon, 3 Feb 2003 12:48:53 -0500
Peter Radcliffe <p...@pir.net> wrote:

PR> Updating to -STABLE as of yesterday morning the cdrom combo drive in
PR> my laptop is not longer recognised.
PR>
PR> ata1-master: ATA identify retries exceeded

Hmm my new Samsung 48x24x48 does that too sometimes at boot, my
- -stable is from 19 Jan. I was blaming the drive ...

It looks like nothing in the ATA code has changed since
December 16 in my sources.

PR> If I use atacontrol to detatch and attach ata1 it appears;

Doh! I was using reinit - that's a much better idea.

- --
C:>WIN | Directable Mirrors
The computer obeys and wins. |A Better Way To Focus The Sun
You lose and Bill collects. | licenses available - see:
| http://www.sohara.org/

------------------------------

End of stable-digest V5 #783
****************************

To Unsubscribe: send mail to majo...@FreeBSD.org
with unsubscribe freebsd-stable-digest in the body of the message

0 new messages