date:20090910


Hi,
i have few IBM scsi drives from old (risc-based) IBM as/400.

These drives have 520KByte sector size instead of 512KByte.
Is there any way how to dump raw data from these drives without
reformat drive to different sector size ?

$ dmesg
...
sd2 at scsibus0 targ 3 lun 0: IBMAS400, DFHSS4W, 4I4I SCSI 2 0/direct 
fixed

sd2: drive offline
...

Thank You for any ideas.

Milan Prihoda

Gov't Fiscal Year End Discounts on SigmaPlot Version 11!

2009-09-10 Thread Systat Software Inc

Gov't
Fiscal Year End 
Discounts on SigmaPlot 11!


Give
  our sales team a call and place your order before
  this offer expires
on Sept 30, 2009!

Systat
  Software Government Sales: (800) 797-7401

Or
  FAX your purchase order to: (800)797-7406

SigmaPlot
11 - The Technical Graphing Standard now
with Complete
Advisory Statistics!

SigmaPlot
  11 now offers a full
range of easy-to-use
  statistical functions
in addition to graphing
  templates and
utilities
  with improved features
for accurate and
  fast data analysis.
Unlike other analysis
  and graphing packages,
spreadsheets or business
  graphing products,
SigmaPlot 11 offers
  complete advisory
statistical analysis
  features along with
a full range of graphing
  templates and
utilities
  for unmatched data
accuracy, speed,
  data analysis and
presentation. SigmaPlot
  is known for its
flexibility to customize
  any object on a
graph
  to meet the exacting
requirements of the
  scientific and
engineering
  community. 
  More information
on SigmaPlot Version 11...


Ask About Our NEW PRODUCTS!


 SigmaCERF,
  The Collaborative Electronic Research Frameworktrade; is
a unique ELN solution designed specifically for
  scientists and RD
organizations. more
  info...

 LISA.lims
  is the PREMIER Laboratory
Information Management
  System. LISA.lims helps you to automate, organize
and standardize complex workflows. more
  info...

 Genomatix
  tools
let you understand the molecular mechanisms
  of gene regulation as a
central part of systems
  biology. Genomatix provides comprehensive,
reliable results. more
  info...

 

TOOLS FOR SCIENCE

Systat Software,
  Inc. 1735 Technology
Drive Suite 430 San
  Jose, CA 95110 USA
Phone: 800-797-7401
  Fax: 800-797-7406

  Email: info-...@systat.com


 
Unsubscribe - 
Our
Web  Email policy
- 
Copyright copy; 2009
Systat
Software, Inc. - All Rights Reserved - 
Please
visit our Website.

Re: IBM 520KByte sector size scsi drives


Jan Stary wrote:

On Sep 10 10:36:41, Milan Prihoda wrote:
  

Hi,
i have few IBM scsi drives from old (risc-based) IBM as/400.

These drives have 520KByte sector size instead of 512KByte.
Is there any way how to dump raw data from these drives without
reformat drive to different sector size ?

$ dmesg
...
sd2 at scsibus0 targ 3 lun 0: IBMAS400, DFHSS4W, 4I4I SCSI 2 0/direct  
fixed

sd2: drive offline
...



What is raw data?
Just dd if=/dev/rsd2c of=somewhere.

Or is there a filesystem you can read/dump?
Then dump | restore as usual.

What are you _really_ trying to do?

  

Yes, just dd if=/dev/rsd2c 

_really_ ? :-) I am just curious :-)

Milan Prihoda

ugly switch problem

2009-09-10 Thread Holger Glaess

hi

i have an running 4.4 firewall cluster on sun x2200 m2 boxes with
additional PCIe quad card with intel chipset.


as switch i use hp 2824 or 3500 procurve witch the latest or close to the
latest firmware.

so, if i do a reboot of the master box i get trouble at the boot time,
the box can't set routing on some interfaces and the firewall rules are
loaded but don't work properly and the whole system is not remote
accessable.

if i do now an pfctl -Fall -f pf.conf anything works fine, just i have to
set the missing static routing entrys.

same time on the switch , he says the link is there , than aways like an
uncontroled flap with an stable link if the box have finished the boot.

now , i do on the box the reload of the rules , everthing ist ok.

what kind of problem is that ?
do the kernel switch off and on the interfaces due the boot time ?

how can i fix this ?

holger

note: i hope it is ok that i attach the long boot output.

[demime 1.01d removed an attachment of type application/octet-stream which had 
a name of 7]

Re: mutt: xterm_title conf var

2009-09-10 Thread Luis Useche

I don't know if this solve your problem or you already did this, but
you need to set the xterm in order to have colors for manuals or vim.
Read from FAQ How do I get color man pages on XTerm?.

Luis



On Thu, Sep 10, 2009 at 3:43 AM, Bernd 'Siggy' Brentrup b...@xlab.net wrote:
 Hi List,

 I'm absolutely new to openbsd so please bear with me if it's not
 the right place to ask my question.  The box I installed 4.5 will
 be running headless so I'm logged in remotely on a color xterm.

 I rsynced mutt configuration from my Ubuntu box because it's also
 running mutt 1.5.18,  most things work out as expected except for
 lack of colors, missing threading and convenient confvars
 xterm_set_titles and xterm_title that I use to display status
 information in the xterm title.

 I'll investigate the first 2 quirks later, as for ther xterm stuff
 my question is did you leave these out on purpose, they are missing
 in your muttrc(5) too (if so why) or should I check Ubuntu's resp.
 Debian's source code to see if they patched mutt.

 Thanks
  Siggy
 --
 O ascii ribbon campaign - stop html mail - www.asciiribbon.org+
 |36 days until|Open Source in Northern Germany: www.free-it.org|
 |www.Ubucon.de|tech contact: bsb-at-free-dash-it-dot-de|
 +--- ceterum censeo javascriptum esse restrictam +

Re: IBM 520KByte sector size scsi drives

2009-09-10 Thread Joachim Schipper

On Thu, Sep 10, 2009 at 12:16:27PM +0200, Milan Prihoda wrote:
 Jan Stary wrote:
 On Sep 10 10:36:41, Milan Prihoda wrote:
 Hi,
 i have few IBM scsi drives from old (risc-based) IBM as/400.
 
 These drives have 520KByte sector size instead of 512KByte.
 Is there any way how to dump raw data from these drives without
 reformat drive to different sector size ?
 
 $ dmesg
 ...
 sd2 at scsibus0 targ 3 lun 0: IBMAS400, DFHSS4W, 4I4I SCSI 2
 0/direct  fixed
 sd2: drive offline
 ...
 
 What is raw data?
 Just dd if=/dev/rsd2c of=somewhere.
 
 Or is there a filesystem you can read/dump?
 Then dump | restore as usual.
 
 What are you _really_ trying to do?
 
 Yes, just dd if=/dev/rsd2c 
 
 _really_ ? :-) I am just curious :-)

dd(1) has an option to set the block size (bs=...). Setting this to a
(multiple of) the block size of the underlying hardware may (or may not)
increase performance.

However, the drivers will take care of any hardware issues and present a
stream of bytes to userland, no matter the block size. So plain dd will
work.

Joachim

Re: mutt: xterm_title conf var

2009-09-10 Thread Joachim Schipper

On Thu, Sep 10, 2009 at 09:43:24AM +0200, Bernd 'Siggy' Brentrup wrote:
 Hi List,
 
 I'm absolutely new to openbsd so please bear with me if it's not
 the right place to ask my question.  The box I installed 4.5 will
 be running headless so I'm logged in remotely on a color xterm.
 
 I rsynced mutt configuration from my Ubuntu box because it's also
 running mutt 1.5.18,  most things work out as expected except for
 lack of colors, missing threading and convenient confvars
 xterm_set_titles and xterm_title that I use to display status
 information in the xterm title.
 
 I'll investigate the first 2 quirks later, as for ther xterm stuff
 my question is did you leave these out on purpose, they are missing
 in your muttrc(5) too (if so why) or should I check Ubuntu's resp.
 Debian's source code to see if they patched mutt.

I'm fairly certain xterm_* is a Debian/Ubuntu improvement: at least,
http://www.mutt.org/doc/devel/manual.html doesn't know about either
option. If it's important to you, you may be able to get the patches
from the Debian package to apply against OpenBSD's port.

Colors do work, with a proper TERM setting (e.g. xterm-xfree86 or
xterm-color; plain xterm doesn't work, since there are things that
identify as xterm but don't handle color.) Try setting TERM properly
and/or running env TERM=xterm-color mutt.

Threading definitely works for me, although you can get some display
issues if the characters sets used don't match. Feel free to ask for the
details; your original post was sufficiently unclear that I cannot guess
your problem.

Joachim

Re: mutt: xterm_title conf var

On Thu, Sep 10, 2009 at 07:21 -0400, Luis Useche wrote:
 I don't know if this solve your problem or you already did this, but
 you need to set the xterm in order to have colors for manuals or vim.
 Read from FAQ How do I get color man pages on XTerm?.

Not reading the FAQ but trying after your hint:

^Z
$ echo $TERM
xterm
$ TERM=xterm-color
$ %
^L

colors appear. I'll fix my .bashrc (resp. .zshrc asa zsh is
installed).

Thanks for your hint
  Siggy
-- 
O ascii ribbon campaign - stop html mail - www.asciiribbon.org+
|36 days until|bsb-at-psycho-dot-informationsanarchistik-dot-de|
|www.Ubucon.de|or:bsb-at-psycho-dot-i21k-dot-de|
+--- ceterum censeo javascriptum esse restrictam +

Brazilian shop gives 404

2009-09-10 Thread Christiano Farina Haesbaert

Brazilian shop gives 404 for www.temporeal.com.br/openbsd

While searching the cds in their system, the last one is 4.4.
I was trying to pre order 4.6 from them, but it seems they won't
have it ?

Re: Supporting OpenBSD

2009-09-10 Thread Daniel Bolgheroni

On Tue, 8 Sep 2009, Nick Holland wrote:

 Thanks to those that contribute money and buy CDs.

I would like to buy CDs, but in Brazil these kind of products have a 
high tax fee applied when they hit the harbour. For a $50 CD, I'll 
probably pay almost $almost $70 to someone I don't want to contribute. 
This doesn't include the shipment cost (~$30 I suspect).

I don't have a Paypal account (yet). If it's worth to trust him, I don't 
know, but I much prefer to donate $50 (although they will deduct 3.9% in 
my case, but at least OpenBSD doesn't have the CD cost) than to pay 
almost the triple to government, shipment, etc. Don't care if I don't 
get the CDs.

Is it possible to OpenBSD to make profit for the project selling books 
or manuals? I don't know the costs or if it's worth (like CDs are better 
for the project than T-shirts, mugs, etc.). It's tax free here, and I 
think: if it's free here, maybe it's somewhere else.

Teers,

--
Daniel Bolgheroni
FEI - Faculdade de Engenharia Industrial
http://www.dbolgheroni.eng.br/mykey

ASCII ribbon campaign ( )
 against HTML e-mail   X
  / \

SOLVED: mutt: xterm_title conf var

On Thu, Sep 10, 2009 at 14:40 +0200, Joachim Schipper wrote:
 On Thu, Sep 10, 2009 at 09:43:24AM +0200, Bernd 'Siggy' Brentrup wrote:

 I'm fairly certain xterm_* is a Debian/Ubuntu improvement: at least,
 http://www.mutt.org/doc/devel/manual.html doesn't know about either
 option. If it's important to you, you may be able to get the patches
 from the Debian package to apply against OpenBSD's port.

If find it quite convenient to see selected status in the xterm title
while reading e.g. a crossposted mail.

 Colors do work, with a proper TERM setting (e.g. xterm-xfree86 or
 xterm-color; plain xterm doesn't work, since there are things that
 identify as xterm but don't handle color.) Try setting TERM properly
 and/or running env TERM=xterm-color mutt.

Sorry you're too late to earn credit on this, Luis pointed me in
the right direction before I saw your post :)

 Threading definitely works for me, although you can get some display
 issues if the characters sets used don't match. Feel free to ask for the
 details; your original post was sufficiently unclear that I cannot guess
 your problem.

I was unclear on purpose because I wanted to investigate these problems
myself.  Solution is: on Debian systems default is 'sort=threads' while
openbsd has 'sort=date'.  After adding 'set sort=threads' to .muttrc
this file now works on all my systems as I want it to.

Thanks to all who responded on list or privately.  Because one of the
private ones stands out by being written in german I'll answer that
one in a separate private mail.

Impressed by the way this thread worked
  Siggy
  
FYI: please visit https://launchpad.net/~bsb
-- 
O ascii ribbon campaign - stop html mail - www.asciiribbon.org+
|36 days until|bsb-at-psycho-dot-informationsanarchistik-dot-de|
|www.Ubucon.de|or:bsb-at-psycho-dot-i21k-dot-de|
+--- ceterum censeo javascriptum esse restrictam +

Re: IBM 520KByte sector size scsi drives

2009-09-10 Thread David Vasek


On Thu, 10 Sep 2009, Joachim Schipper wrote:


On Thu, Sep 10, 2009 at 12:16:27PM +0200, Milan Prihoda wrote:

Jan Stary wrote:

On Sep 10 10:36:41, Milan Prihoda wrote:

Hi,
i have few IBM scsi drives from old (risc-based) IBM as/400.

These drives have 520KByte sector size instead of 512KByte.
Is there any way how to dump raw data from these drives without
reformat drive to different sector size ?

$ dmesg
...
sd2 at scsibus0 targ 3 lun 0: IBMAS400, DFHSS4W, 4I4I SCSI 2
0/direct  fixed
sd2: drive offline
...


What is raw data?
Just dd if=/dev/rsd2c of=somewhere.

Or is there a filesystem you can read/dump?
Then dump | restore as usual.

What are you _really_ trying to do?


Yes, just dd if=/dev/rsd2c 

_really_ ? :-) I am just curious :-)


dd(1) has an option to set the block size (bs=...). Setting this to a
(multiple of) the block size of the underlying hardware may (or may not)
increase performance.

However, the drivers will take care of any hardware issues and present a
stream of bytes to userland, no matter the block size. So plain dd will
work.


The drive is reported as offline.

As I understand it from sys/scsi/sd.c, drive offline means an error was 
reported by scsi_test_unit_ready() function call. Possibly the 520 
bytes/sector is not compatible with the controller the OP used or with the 
OpenBSD driver/scsi subsystem? Don't know. I doubt that media (i.e. 
blocks) is accessible in this condition.


Regards,
David

Re: IBM 520KByte sector size scsi drives

David Vasek wrote:
 The drive is reported as offline.

 As I understand it from sys/scsi/sd.c, drive offline means an error was
 reported by scsi_test_unit_ready() function call. Possibly the 520
 bytes/sector is not compatible with the controller the OP used or with the
 OpenBSD driver/scsi subsystem? Don't know. I doubt that media (i.e.
 blocks) is accessible in this condition.

 Regards,
 David

Hi,

I have an ATAPI LS-120 drive that always reports itself as being
offline, but it functions properly.. it would however be nice if the
device reported such online/offline events, but that could be a
limitation of the hardware.

sd0 at scsibus0 targ 0 lun 0: MATSHITA, LS-120 VER5 00, F527 ATAPI
0/direct removable
sd0: drive offline

-Brynet

Re: IBM 520KByte sector size scsi drives

Milan Prihoda wrote:
 These drives have 520KByte sector size instead of 512KByte.
 Is there any way how to dump raw data from these drives without
 reformat drive to different sector size ?

Most disks have a sector size of 512 bytes, not kilobytes.

-Brynet

Re: OpenBSD hacking

2009-09-10 Thread Luis Useche

Thank you all for your suggestions. I will respond by names.

Jesus: I sent this email to this list because is the only contact I
know to the OBSD developers. Are there other list more appropriate for
my question?

Claudio: I don't think FS2 assumptions are completely out-dated. If
this were true, many of the file systems and I/O schedulers
optimizations based on locality would not work (I guess this is a
discussion for another thread). I thought about the Journaling for FFS
before but I felt it was too complicated for an initial kernel
project; maybe I'm wrong. In any case, I will find out more about
this. Thanks.

David: Sure, it does not matter if it is not FS2 related. We can talk
by email or through the misc list, whichever you prefer.

Edd: I haven't seen anything like a snapshot in FFS. I think snapshots
come really unnatural to FFS-like file systems. This is usually
implemented in COW file systems. For this reason, I was also thinking
on porting LFS from NetBSD and include the snapshoting functionality.
But again, I thought this could probably be somewhat complicated. It
could be a future project though.

Jonathan: Again, this is journaling for FFS. I guess same comments for
Claudio apply to you.

Thanks,

Luis

Amanda backup problem on OpenBSD 4.5

2009-09-10 Thread Michael Burk

Hi All,

We have been working for a couple weeks with the Amanda developers to get
their latest release working on OpenBSD 4.5. Amanda stopped working with
OpenBSD somewhere around version 2.5.1 (3 years ago). We've been focusing on
a failure in the dump process on the client.

On the client side, dump(8) is started with its output to a blocking pipe.
However, dump fails with EAGAIN, which we understand should be impossible.
One of the developers outlined the basic process as follows:

In amandad process:
 pipe(pipefd)
 dup2(pipefd[1],b)
 fork
 in the child:
exec sendbackup process
dup2(b,c)
write c #fail with EAGAIN

We added some debug messages to verify that the pipe was blocking. In so
doing, we discovered that the debug messages fixed the problem, and the
dump succeeded. Narrowing this down, we found the absolute minimal patch was
simply inserting the following line before the dup2 above:
 fcntl(datafd, F_GETFL, 0);

This adds to the mystery; why would doing a call that simply returns a flag
change the behavior of the program? Is there some side effect of this call?

We have tried this on several machines (at least the sparc64, amd64
platforms), all running 4.5. dump is invoked as follows:
/sbin/dump dump 0usf 1048576 - /dev/rsd0a

I don't know what other details might be helpful; let me know and I'll send
them.

Thank you for any insight.

-- Michael

Re: Amanda backup problem on OpenBSD 4.5

2009-09-10 Thread Stuart Henderson

On 2009-09-10, Michael Burk bur...@gmail.com wrote:
 We have been working for a couple weeks with the Amanda developers to get
 their latest release working on OpenBSD 4.5. Amanda stopped working with
 OpenBSD somewhere around version 2.5.1 (3 years ago). We've been focusing on
 a failure in the dump process on the client.

Does newer Amanda use threads? With pthread, stdio will be silently changed
to _non_ blocking.



 On the client side, dump(8) is started with its output to a blocking pipe.
 However, dump fails with EAGAIN, which we understand should be impossible.
 One of the developers outlined the basic process as follows:

 In amandad process:
  pipe(pipefd)
  dup2(pipefd[1],b)
  fork
  in the child:
 exec sendbackup process
 dup2(b,c)
 write c #fail with EAGAIN

 We added some debug messages to verify that the pipe was blocking. In so
 doing, we discovered that the debug messages fixed the problem, and the
 dump succeeded. Narrowing this down, we found the absolute minimal patch was
 simply inserting the following line before the dup2 above:
  fcntl(datafd, F_GETFL, 0);

 This adds to the mystery; why would doing a call that simply returns a flag
 change the behavior of the program? Is there some side effect of this call?

 We have tried this on several machines (at least the sparc64, amd64
 platforms), all running 4.5. dump is invoked as follows:
 /sbin/dump dump 0usf 1048576 - /dev/rsd0a

 I don't know what other details might be helpful; let me know and I'll send
 them.

 Thank you for any insight.

 -- Michael

Re: Amanda backup problem on OpenBSD 4.5

2009-09-10 Thread Michael Burk

It links with libpthread. I am told that on the client side it uses a single
thread. Would that make a difference?

Also, I forgot to include a link to the *very* long thread on the Amanda
list discussing this problem:

http://archives.zmanda.com/amanda-archives/viewtopic.php?t=4763sid=f86451987866cfccc9fe363fe7546a7b

Thanks,
Michael

On Thu, Sep 10, 2009 at 10:09 AM, Stuart Henderson s...@spacehopper.orgwrote:

On 2009-09-10, Michael Burk bur...@gmail.com wrote:
We have been working for a couple weeks with the Amanda developers to get
their latest release working on OpenBSD 4.5. Amanda stopped working with
OpenBSD somewhere around version 2.5.1 (3 years ago). We've been focusing
on
a failure in the dump process on the client.

Does newer Amanda use threads? With pthread, stdio will be silently changed
to _non_ blocking.

On the client side, dump(8) is started with its output to a blocking
pipe.
However, dump fails with EAGAIN, which we understand should be
impossible.
One of the developers outlined the basic process as follows:

In amandad process:
pipe(pipefd)
dup2(pipefd[1],b)
fork
in the child:
exec sendbackup process
dup2(b,c)
write c #fail with EAGAIN

We added some debug messages to verify that the pipe was blocking. In so
doing, we discovered that the debug messages fixed the problem, and the
dump succeeded. Narrowing this down, we found the absolute minimal patch
was
simply inserting the following line before the dup2 above:
fcntl(datafd, F_GETFL, 0);

This adds to the mystery; why would doing a call that simply returns a
flag
change the behavior of the program? Is there some side effect of this
call?

We have tried this on several machines (at least the sparc64, amd64
platforms), all running 4.5. dump is invoked as follows:
/sbin/dump dump 0usf 1048576 - /dev/rsd0a

I don't know what other details might be helpful; let me know and I'll
send
them.

Thank you for any insight.

-- Michael

Re: IBM 520KByte sector size scsi drives


Brynet wrote:

Milan Prihoda wrote:
  

These drives have 520KByte sector size instead of 512KByte.
Is there any way how to dump raw data from these drives without
reformat drive to different sector size ?



Most disks have a sector size of 512 bytes, not kilobytes.

-Brynet

  

Hi,
i should apologize for incompletely question (with mistakes)
so again (and better :-)

Yes, of course drives have sector size of 520bytes not kilobytes,
it was my typo fail.


# dmesg | grep sd2
sd2 at scsibus0 targ 2 lun 0: IBMAS400, DFHSS4W, 4I4I SCSI2 0/direct fixed
sd2: drive offline

# fdisk sd2
fdisk: sd2: Device not configured

# disklabel sd2
disklabel: /dev/rsd2c: Device not configured

# dd=/dev/rsd2c of=disk1.img
dd: /dev/rsd2c: Device not configured



I tried this drive in a linux box and dmesg told me:
# dmesg | grep sdb
sd 0:0:3:0: [sdb] Unsupported sector size 520.
sd 0:0:3:0: [sdb] 0 512-byte hardware sectors: (0 B/0 B)
sd 0:0:3:0: [sdb] Write Protect is off
sd 0:0:3:0: [sdb] Mode Sense: a7 00 10 08
sd 0:0:3:0: [sdb] Write cache: disabled, read cache: enabled, supports 
DPO and FUA

sd 0:0:3:0: [sdb] Attached SCSI disk

and there i get knowledge about 520byte sector size.

I would like to get raw data from this drive.
Ok - if it's impossible, i want to say it. :-)

Thank. Milan

Re: Amanda backup problem on OpenBSD 4.5

2009-09-10 Thread Michael Burk

If pthreads is the issue, then I have two questions:
1) is the behavior of changing stdio to non_blocking when using pthreads
unique to OpenBSD?
2) why would this call fix the problem:
fcntl(datafd, F_GETFL, 0);
does that have other side effects?

Thanks,
Michael

On Thu, Sep 10, 2009 at 10:44 AM, Bob Beck b...@ualberta.ca wrote:

That is almost certainly the problem. Ditch the pthread dependancy and it
will probably work fine.

On Thu, Sep 10, 2009 at 10:34 AM, Michael Burk bur...@gmail.com wrote:

It links with libpthread. I am told that on the client side it uses a
single
thread. Would that make a difference?

Also, I forgot to include a link to the *very* long thread on the Amanda
list discussing this problem:

http://archives.zmanda.com/amanda-archives/viewtopic.php?t=4763sid=f86451987866cfccc9fe363fe7546a7b

Thanks,
Michael

On Thu, Sep 10, 2009 at 10:09 AM, Stuart Henderson s...@spacehopper.org
wrote:

On 2009-09-10, Michael Burk bur...@gmail.com wrote:
We have been working for a couple weeks with the Amanda developers to
get
their latest release working on OpenBSD 4.5. Amanda stopped working
with
OpenBSD somewhere around version 2.5.1 (3 years ago). We've been
focusing
on
a failure in the dump process on the client.

Does newer Amanda use threads? With pthread, stdio will be silently
changed
to _non_ blocking.

In amandad process:
pipe(pipefd)
dup2(pipefd[1],b)
fork
in the child:
exec sendbackup process
dup2(b,c)
write c #fail with EAGAIN

We added some debug messages to verify that the pipe was blocking. In
so
doing, we discovered that the debug messages fixed the problem, and
the
dump succeeded. Narrowing this down, we found the absolute minimal
patch
was
simply inserting the following line before the dup2 above:
fcntl(datafd, F_GETFL, 0);

This adds to the mystery; why would doing a call that simply returns a
flag
change the behavior of the program? Is there some side effect of this
call?

We have tried this on several machines (at least the sparc64, amd64
platforms), all running 4.5. dump is invoked as follows:
/sbin/dump dump 0usf 1048576 - /dev/rsd0a

I don't know what other details might be helpful; let me know and I'll
send
them.

Thank you for any insight.

-- Michael

Re: Amanda backup problem on OpenBSD 4.5

2009-09-10 Thread Stuart Henderson

On 2009-09-10, Michael Burk bur...@gmail.com wrote:
 It links with libpthread.

 I am told that on the client side it uses a single thread.
 Would that make a difference?

That is exactly the problem then. Just linking with libpthread is enough,
for userland threading to work it has to use non-blocking FDs and wraps
libc functions to do this.

 Also, I forgot to include a link to the *very* long thread on the Amanda
 list discussing this problem:

 http://archives.zmanda.com/amanda-archives/viewtopic.php?t=4763sid=f86451987866cfccc9fe363fe7546a7b

ouch. yes, *very* long.

Re: Amanda backup problem on OpenBSD 4.5

2009-09-10 Thread Håkan Kvist

-BEGIN PGP SIGNED MESSAGE-
Hash: SHA1

Michael Burk wrote the following lines on 09/10/2009 06:58 PM:
 If pthreads is the issue, then I have two questions:
 1) is the behavior of changing stdio to non_blocking when using pthreads
 unique to OpenBSD?
 2) why would this call fix the problem:
 fcntl(datafd, F_GETFL, 0);
 does that have other side effects?
 
 Thanks,
 Michael

Is guess it is the same as for older versions of FreeBSD, I googled link:
http://lists.freebsd.org/pipermail/freebsd-bugs/2003-May/000615.html

I assume the same thing applies to OpenBSD:
The threading takes place in userspace in OpenBSD and the kernel does
not know anything about threads.

When doing a blocking read in a threaded application, all threads would
be blocked since the only thing the kernel knows about is the process.

...
 list discussing this problem:

http://archives.zmanda.com/amanda-archives/viewtopic.php?t=4763sid=f86451987866cfccc9fe363fe7546a7b

If I got it correctly the clients only have one thread, so then I guess
you could do:
 fcntl(datafd, F_GETFL, 0);
without getting any side effects.

But on the other hand that is an ugly work around, and ugly work arounds
 tends to cause other problems in the future. So I say get rid of the
pthreads dependency since the clients are single threaded anyway.

br
Hekan
Comment: Using GnuPG with Mozilla - http://enigmail.mozdev.org

iD8DBQFKqTVLfu+aNIkyxzsRAuzuAJ4wP4qsamRwxoNGqW4EhJyz8AAuFACeLE7Y
l4Q1AvYMn3Uh/nMELi/gRTs=
=88yz
-END PGP SIGNATURE-

Re: IBM 520KByte sector size scsi drives

Hi Milan,

David Vasek was on the right track with his message, it seems likely
that the driver is bailing out on line 1439 in /usr/src/sys/scsi/sd.c.

1423:   /*
1424:* Restrict blksize values to powers of two between 512 and 64k.
1425:*/
1426:   switch (dp-blksize) {
1427:   case 0x200: /* == 512, == DEV_BSIZE on all architectures. */
1428:   case 0x400:
1429:   case 0x800:
1430:   case 0x1000:
1431:   case 0x2000:
1432:   case 0x4000:
1433:   case 0x8000:
1434:   case 0x1:
1435:   break;
1436:   default:
1437:   SC_DEBUG(sd-sc_link, SDEV_DB1,
1438:   (sd_get_parms: bad blksize: %#x\n, dp-blksize));
1439:   return (SDGP_RESULT_OFFLINE);
1440:   }

Adding a special case for 0x208 after 0x200 may let the driver
attach.. this way you can dump the raw data off of the disk.

All bets are off after dumping the data, what file system were you using on it?

-Brynet

Informacion importante nuevos horarios de atencion

2009-09-10 Thread Computerland

 http://www.computerlandnews.info/lt.php?id=ZR4HBwYNCQ9cCEUGA0hXVAoKUQAF

 Ofertas
http://www.computerlandnews.info/lt.php?id=ZR4HBwYNCQ9cC0UGA0hXVAoKUQAF

 Noticias
http://www.computerlandnews.info/lt.php?id=ZR4HBwYNCQ9cCkUGA0hXVAoKUQAF

 Novedades
http://www.computerlandnews.info/lt.php?id=ZR4HBwYNCQ9cDUUGA0hXVAoKUQAF

 

 http://www.computerlandnews.info/lt.php?id=ZR4HBwYNCQ9cDEUGA0hXVAoKUQAF

 http://www.computerlandnews.info/lt.php?id=ZR4HBwYNCQ9cC0UGA0hXVAoKUQAF

 mas informacion
http://www.computerlandnews.info/lt.php?id=ZR4HBwYNCQ9cDEUGA0hXVAoKUQAF

 mas informacion
http://www.computerlandnews.info/lt.php?id=ZR4HBwYNCQ9cC0UGA0hXVAoKUQAF

 http://www.computerlandnews.info/lt.php?id=ZR4HBwYNCQ9cC0UGA0hXVAoKUQAF

 http://www.computerlandnews.info/lt.php?id=ZR4HBwYNCQ9cC0UGA0hXVAoKUQAF

 mas informacion
http://www.computerlandnews.info/lt.php?id=ZR4HBwYNCQ9cC0UGA0hXVAoKUQAF

 mas informacion
http://www.computerlandnews.info/lt.php?id=ZR4HBwYNCQ9cC0UGA0hXVAoKUQAF

 http://www.computerlandnews.info/lt.php?id=ZR4HBwYNCQ9cC0UGA0hXVAoKUQAF

 http://www.computerlandnews.info/lt.php?id=ZR4HBwYNCQ9cC0UGA0hXVAoKUQAF

 mas informacion
http://www.computerlandnews.info/lt.php?id=ZR4HBwYNCQ9cC0UGA0hXVAoKUQAF

 mas informacion
http://www.computerlandnews.info/lt.php?id=ZR4HBwYNCQ9cC0UGA0hXVAoKUQAF

 http://www.computerlandnews.info/lt.php?id=ZR4HBwYNCQ9cC0UGA0hXVAoKUQAF

 http://www.computerlandnews.info/lt.php?id=ZR4HBwYNCQ9cD0UGA0hXVAoKUQAF

 **GUAYMAS

 Calle 10 No. 145 A

 Col. Centro

 Guaymas, Sonora

 Tel: 01(622) 224-6277**

 **COMPUTERLAND MATRIZ

 Gabriel CastaC1os No. 9

 Col. Arcos Vallarta

 Guadalajara Jalisco

 Tel: 01(33) 3915-**

 **CIUDAD DE MEXICO

 Av. IPN No. 2051 Piso 1

 Col. Linda Vista

 MC)xico, D.F.

 Tel: 01(55) 4334-0344**



center
Para darse de baja de esta lista visite
http://www.computerlandnews.info/lt.php?id=ZR4HBwYNCQ5VC0UGA0hXVAoKUQAF

Para actualizar sus preferencias visite
http://www.computerlandnews.info/lt.php?id=ZR4HBwYNCQ9cAUUGA0hXVAoKUQAF

Para reenviar esto a un amigo visite
http://www.computerlandnews.info/lt.php?id=ZR4HBwYNCQ9cAEUGA0hXVAoKUQAF

a
href='http://www.computerlandnews.info/lt.php?id=ZR4HBwYNCQ5VCkUGA0hXVAoKUQAFPolmtica
De Privacidad/a



/center


--
Powered by Wadil Works, www.wadilworks.com --

Re: IBM 520KByte sector size scsi drives

2009-09-10 Thread Marco Peereboom

I would love to get my hands on such a drive :-)

EMC uses the same magic; they cram stuff in the additional 8 bytes.

On Thu, Sep 10, 2009 at 06:41:08PM +0200, Milan Prihoda wrote:
 Brynet wrote:
 Milan Prihoda wrote:
   
 These drives have 520KByte sector size instead of 512KByte.
 Is there any way how to dump raw data from these drives without
 reformat drive to different sector size ?
 

 Most disks have a sector size of 512 bytes, not kilobytes.

 -Brynet

   
 Hi,
 i should apologize for incompletely question (with mistakes)
 so again (and better :-)

 Yes, of course drives have sector size of 520bytes not kilobytes,
 it was my typo fail.


 # dmesg | grep sd2
 sd2 at scsibus0 targ 2 lun 0: IBMAS400, DFHSS4W, 4I4I SCSI2 0/direct fixed
 sd2: drive offline

 # fdisk sd2
 fdisk: sd2: Device not configured

 # disklabel sd2
 disklabel: /dev/rsd2c: Device not configured

 # dd=/dev/rsd2c of=disk1.img
 dd: /dev/rsd2c: Device not configured



 I tried this drive in a linux box and dmesg told me:
 # dmesg | grep sdb
 sd 0:0:3:0: [sdb] Unsupported sector size 520.
 sd 0:0:3:0: [sdb] 0 512-byte hardware sectors: (0 B/0 B)
 sd 0:0:3:0: [sdb] Write Protect is off
 sd 0:0:3:0: [sdb] Mode Sense: a7 00 10 08
 sd 0:0:3:0: [sdb] Write cache: disabled, read cache: enabled, supports  
 DPO and FUA
 sd 0:0:3:0: [sdb] Attached SCSI disk

 and there i get knowledge about 520byte sector size.

 I would like to get raw data from this drive.
 Ok - if it's impossible, i want to say it. :-)

 Thank. Milan

Re: Amanda backup problem on OpenBSD 4.5

2009-09-10 Thread Dustin J. Mitchell

On Thu, Sep 10, 2009 at 12:34 PM, Michael Burkbur...@gmail.com wrote:
 Does newer Amanda use threads? With pthread, stdio will be silently
 changed to _non_ blocking.

Wait, what?!

When does this happen?  At exec? And does it just apply to fd's 0-2?
Can you point me to a manpage?

Dustin

(michael and stuart - if this doesn't make it to misc@, please forward it along)

-- 
Open Source Storage Engineer
http://www.zmanda.com

Re: Amanda backup problem on OpenBSD 4.5

2009-09-10 Thread Philip Guenther

On Thu, Sep 10, 2009 at 11:13 AM, Dustin J. Mitchell dus...@zmanda.com
wrote:
 On Thu, Sep 10, 2009 at 12:34 PM, Michael Burkbur...@gmail.com wrote:
 Does newer Amanda use threads? With pthread, stdio will be silently
 changed to _non_ blocking.

 Wait, what?!

This is all behind the applications back, between the thread library
and the kernel.  Ideally it would be completely invisible to the
application, but while the library can hide it from and single
process, it can't hide it from other processes and there are problems
with the threaded process forks and exits or execs.


 When does this happen?  At exec? And does it just apply to fd's 0-2?
 Can you point me to a manpage?

It happens at initialization time of libpthread and is undone at exit
and exec.  It applies to all fds.  There is no manpage, as it's not
(supposed to be) application visible.


Philip Guenther

OnlineHotel, 5 years working with you for your success!

2009-09-10 Thread OnlineHotel

Online Hotel

 

OnlineHotel, 5 years working with you for your success!

 

Hello

 

Meet one of the best websites of Latin America on online hotel reservations.
Having 5 years of experience on the market, OnlineHotel brings out its
newest reservation system.

 

OnlineHotel system helps you sell daily reservations online to the whole
world with instant confirmation and much more safety. Besides, our control
and transaction systems are entirely developed with the most modern
techniques on the digital market.

 

See our advantages:

 

- Free registration and publicity, there is no service taxes;

 

- By registering OnlineHotel your hotel gets access to an exclusive Control
Panel. You can manage price, vacancy and more;

 

- Guests pay OnlineHotel just 10% of the daily reservation. The other 90% is
paid directly to your hotel according to your payment policy;

 

- OnlineHotel gives you access to an exclusive Control Panel to monitor your
reservation requests and necessary configurations (price, vacancy, pictures,
etc);

 

- When a reservation request is confirmed through OnlineHotel, you receive a
message in your registered e-mail. You can also follow these information on
your Control Panel;

 

- You will receive reservation requests from all over the world!

 

Meet right now these news tools at
http://onlinehotel.com.br/secure/hotel_register
http://onlinehotel.com.br/secure/hotel_register and be a part of this world.

 

Thank you for you attention. We look forward to see you with us!

 

Sincerely,

 

 

OnlineHotel

 http://www.onlinehotel.com.br/ http://www.onlinehotel.com.br

[demime 1.01d removed an attachment of type image/jpeg which had a name of 
image001.jpg]

Re: IBM 520KByte sector size scsi drives


Hi Brynet,
thank You very much for Your idea.
It works :-)

Brynet wrote:

Hi Milan,

David Vasek was on the right track with his message, it seems likely
that the driver is bailing out on line 1439 in /usr/src/sys/scsi/sd.c.

1423:   /*
1424:* Restrict blksize values to powers of two between 512 and 64k.
1425:*/
1426:   switch (dp-blksize) {
1427:   case 0x200: /* == 512, == DEV_BSIZE on all architectures. */
1428:   case 0x400:
1429:   case 0x800:
1430:   case 0x1000:
1431:   case 0x2000:
1432:   case 0x4000:
1433:   case 0x8000:
1434:   case 0x1:
1435:   break;
1436:   default:
1437:   SC_DEBUG(sd-sc_link, SDEV_DB1,
1438:   (sd_get_parms: bad blksize: %#x\n, dp-blksize));
1439:   return (SDGP_RESULT_OFFLINE);
1440:   }

Adding a special case for 0x208 after 0x200 may let the driver
attach.. this way you can dump the raw data off of the disk.

All bets are off after dumping the data, what file system were you using on it?

-Brynet

Re: IBM 520KByte sector size scsi drives


Milan Prihoda wrote:

Hi Brynet,
thank You very much for Your idea.
It works :-)

Brynet wrote:

Hi Milan,

David Vasek was on the right track with his message, it seems likely
that the driver is bailing out on line 1439 in /usr/src/sys/scsi/sd.c.

1423:/*
1424: * Restrict blksize values to powers of two between 512 and 
64k.

1425: */
1426:switch (dp-blksize) {
1427:case 0x200: /* == 512, == DEV_BSIZE on all 
architectures. */

1428:case 0x400:
1429:case 0x800:
1430:case 0x1000:
1431:case 0x2000:
1432:case 0x4000:
1433:case 0x8000:
1434:case 0x1:
1435:break;
1436:default:
1437:SC_DEBUG(sd-sc_link, SDEV_DB1,
1438:(sd_get_parms: bad blksize: %#x\n, dp-blksize));
1439:return (SDGP_RESULT_OFFLINE);
1440:}

Adding a special case for 0x208 after 0x200 may let the driver
attach.. this way you can dump the raw data off of the disk.

All bets are off after dumping the data, what file system were you 
using on it?

I have no idea about filesystem, because these drives are from old
(but risc) as/400 machine. There are four root drives, i dont know, 
in which configuration of raid (if there is smth like raid) they are.

I am just curious.
So again, thank you all.


-Brynet

Re: IBM 520KByte sector size scsi drives


Brynet wrote:

Hi Milan,

I'm glad that it works.. can you post the new attachment message? :)

-Brynet

  

Do you mean this :?

# dmesg | grep sd2
sd2 at scsibus0 targ 2 lun 0: IBMAS400, DFHSS4W, 4I4I SCSI2 0/direct fixed
sd2: 4173MB, 520 bytes/sec, 8414494 sec total

Re: IBM 520KByte sector size scsi drives

Hi Milan,

I'm glad that it works.. can you post the new attachment message? :)

-Brynet

Re: Amanda backup problem on OpenBSD 4.5

2009-09-10 Thread Dustin J. Mitchell

On Thu, Sep 10, 2009 at 2:29 PM, Philip Guentherguent...@gmail.com wrote:
 It happens at initialization time of libpthread and is undone at exit
 and exec. B It applies to all fds. B There is no manpage, as it's not
 (supposed to be) application visible.

OK, so basically this is an OpenBSD/pthread bug.  I'm still sort of
shell-shocked here.  The idea is that *all* file descriptors are made
non-blocking, and then the file-descriptor syscalls are wrapped to
pretend they are blocking.  As long as all of the code operating on
those pseudo-blocking files is using the wrapped syscalls, all is
well.  However, when one of those files is accessed by a binary not
linked to libpthread, the illusion breaks down.

Do I have that right?

Why does the fcntl(fd, F_GETFL) fix it?  Does it really fix it, or are
we just lucky?

This sounds like more of a design bug than a code bug, so I don't
expect a fix in a 4.5 patch release.  So we'll need to do something to
work around it in Amanda.  Presumably this is a fairly common problem,
so what is the usual solution?

Dustin

--
Open Source Storage Engineer
http://www.zmanda.com

Re: IBM 520KByte sector size scsi drives


Line 1423 and following of /usr/src/sys/scsi/sd.c show that we will
only accept devices with sector sizes that are multiples of 512
(DEV_BSIZE). Others will be impossible to open.

Now, if you delete that code you may be able to open the raw device.
No idea, but I'd be interested in what happens. :-).

DEV_BSIZE of 512 is deeply imbedded in the kernel.

 Ken

  


Hi Ken,
I not deleted any line

# diff -u sd.c-origin sd.c 
--- sd.c-origin Thu Sep 10 21:53:44 2009

+++ sd.cThu Sep 10 20:17:41 2009
@@ -1425,6 +1425,7 @@
*/
   switch (dp-blksize) {
   case 0x200: /* == 512, == DEV_BSIZE on all architectures. */
+   case 0x208:
   case 0x400:
   case 0x800:
   case 0x1000:


and now,
# dd if=/dev/rsd2c of=disk1.img bs=520

works for me :-)

Relevant article

2009-09-10 Thread Daniel Bolgheroni

Hi,

since Nick Holland touched on the DESIGN aspect in his e-mail regarding 
supporting OpenBSD, I think this article pretty much reflects the 
reality here.

http://www.osnews.com/story/22135/The_Problem_with_Design_and_Implementation

Just think it's worth reading.

Teers,

--
Daniel Bolgheroni
FEI - Faculdade de Engenharia Industrial
http://www.dbolgheroni.eng.br/mykey

ASCII ribbon campaign ( )
 against HTML e-mail   X
  / \

Re: IBM 520KByte sector size scsi drives

2009-09-10 Thread Kenneth R Westerback

On Thu, Sep 10, 2009 at 06:41:08PM +0200, Milan Prihoda wrote:
 Brynet wrote:
 Milan Prihoda wrote:
 These drives have 520KByte sector size instead of 512KByte.
 Is there any way how to dump raw data from these drives without
 reformat drive to different sector size ?
 
 Most disks have a sector size of 512 bytes, not kilobytes.
 
 -Brynet
 
 Hi,
 i should apologize for incompletely question (with mistakes)
 so again (and better :-)
 
 Yes, of course drives have sector size of 520bytes not kilobytes,
 it was my typo fail.
 
 
 # dmesg | grep sd2
 sd2 at scsibus0 targ 2 lun 0: IBMAS400, DFHSS4W, 4I4I SCSI2 0/direct fixed
 sd2: drive offline

Line 1423 and following of /usr/src/sys/scsi/sd.c show that we will
only accept devices with sector sizes that are multiples of 512
(DEV_BSIZE). Others will be impossible to open.

Now, if you delete that code you may be able to open the raw device.
No idea, but I'd be interested in what happens. :-).

DEV_BSIZE of 512 is deeply imbedded in the kernel.

 Ken

 
 # fdisk sd2
 fdisk: sd2: Device not configured
 
 # disklabel sd2
 disklabel: /dev/rsd2c: Device not configured
 
 # dd=/dev/rsd2c of=disk1.img
 dd: /dev/rsd2c: Device not configured
 
 
 
 I tried this drive in a linux box and dmesg told me:
 # dmesg | grep sdb
 sd 0:0:3:0: [sdb] Unsupported sector size 520.
 sd 0:0:3:0: [sdb] 0 512-byte hardware sectors: (0 B/0 B)
 sd 0:0:3:0: [sdb] Write Protect is off
 sd 0:0:3:0: [sdb] Mode Sense: a7 00 10 08
 sd 0:0:3:0: [sdb] Write cache: disabled, read cache: enabled,
 supports DPO and FUA
 sd 0:0:3:0: [sdb] Attached SCSI disk
 
 and there i get knowledge about 520byte sector size.
 
 I would like to get raw data from this drive.
 Ok - if it's impossible, i want to say it. :-)
 
 Thank. Milan

Re: Relevant article

On Thu, Sep 10, 2009 at 17:09 +, Daniel Bolgheroni wrote:
 Hi,

 since Nick Holland touched on the DESIGN aspect in his e-mail regarding
 supporting OpenBSD, I think this article pretty much reflects the
 reality here.


http://www.osnews.com/story/22135/The_Problem_with_Design_and_Implementation

 Just think it's worth reading.

What came to my mind while reading this article is an almost 30yo
tale from the times of my first encounter with Unix release 6 on
PDP11.  I was studying CS at TU Berlin then and we had a OS Prof
Siggi Schindler (he's the reason why I use the 'y') who also
led a course on the ISO 7 layer model with practical applications.
AFAIR everything essentially boiled down to having a *precise
specification language*.  If you have such a language the only
thing left to do is building a machine that accepts this language
as it's programming language.  In this sense implementing a
specification in any existing programming language is equivalent
to implementing a subset of said machine on top of an existing
one.

Siggy
--
O ascii ribbon campaign - stop html mail - www.asciiribbon.org+
|36 days until|bsb-at-psycho-dot-informationsanarchistik-dot-de|
|www.Ubucon.de|or:bsb-at-psycho-dot-i21k-dot-de|
+--- ceterum censeo javascriptum esse restrictam +

[demime 1.01d removed an attachment of type application/pgp-signature which had 
a name of signature.asc]

Re: IBM 520KByte sector size scsi drives

2009-09-10 Thread Marco Peereboom

There are a million spots where 512 is hard coded...

On Thu, Sep 10, 2009 at 10:03:07PM +0200, Milan Prihoda wrote:
 Line 1423 and following of /usr/src/sys/scsi/sd.c show that we will
 only accept devices with sector sizes that are multiples of 512
 (DEV_BSIZE). Others will be impossible to open.

 Now, if you delete that code you may be able to open the raw device.
 No idea, but I'd be interested in what happens. :-).

 DEV_BSIZE of 512 is deeply imbedded in the kernel.

  Ken

   

 Hi Ken,
 I not deleted any line

 # diff -u sd.c-origin sd.c --- sd.c-origin Thu Sep 10 21:53:44 2009
 +++ sd.cThu Sep 10 20:17:41 2009
 @@ -1425,6 +1425,7 @@
 */
switch (dp-blksize) {
case 0x200: /* == 512, == DEV_BSIZE on all architectures. */
 +   case 0x208:
case 0x400:
case 0x800:
case 0x1000:


 and now,
 # dd if=/dev/rsd2c of=disk1.img bs=520

 works for me :-)

Re: Relevant article