Re: [9fans] aoe on sources

Re: [9fans] aoe on sources

[YAMANASHI Takeshi]

Hi, erik, thanks for the new vblade.  Now i can proceed to
try gPXE booting a linux from the vblade.

> that's one of the first things to look for.  the other is that the ctl file
> "flags:" line should contain the string "up".

'ctl' contained this with the old vblade.

state: up
nopen: 1
nout: 0
nmaxout: 24
nframes: 24
maxbcnt: 1024
fw: 0913
model: Plan 9 Vblade
serial: 
firmware: 2
flag: llba up jumbo 

I put the old vblade at:
	http://www.tip9ug.jp/who/nashi/vblade.old

-- 

Re: [9fans] aoe on sources

[erik quanstrom]

> Hi, erik,
> 
> I got the AoE target mounted using sdata successfuly.
> I recompiled vblade from sources and that was it.

great.  thanks for the bug report.

> with the old vblade, it reported that 'data' file is zero lengthed.
> 
>   term% ls -l /dev/aoe/0.0
>   --rw-r--r-- a 0 glenda glenda   0 Aug 16 12:06 /dev/aoe/0.0/config
>   --rw-r--r-- a 0 glenda glenda   0 Aug 16 12:06 /dev/aoe/0.0/ctl
>   --rw-r----- a 0 glenda glenda   0 Aug 16 12:06 /dev/aoe/0.0/data
>   d-r-xr-xr-x a 0 glenda glenda   0 Aug 16 12:06 /dev/aoe/0.0/devlink
>   --rw-r--r-- a 0 glenda glenda 512 Aug 16 12:06 /dev/aoe/0.0/ident

that's one of the first things to look for.  the other is that the ctl file
"flags:" line should contain the string "up".

- erik

Re: [9fans] aoe on sources

[YAMANASHI Takeshi]

Hi, erik,

I got the AoE target mounted using sdata successfuly.
I recompiled vblade from sources and that was it.


> > mounted the target with:
> >     term% bind -b '#ae' /dev
> 
> i think you mean '#?'?

Yes. I wasn't sure if the character would go out with my mailer.


> and check the reported size of the target with

with the old vblade, it reported that 'data' file is zero lengthed.

  term% ls -l /dev/aoe/0.0
  --rw-r--r-- a 0 glenda glenda   0 Aug 16 12:06 /dev/aoe/0.0/config
  --rw-r--r-- a 0 glenda glenda   0 Aug 16 12:06 /dev/aoe/0.0/ctl
  --rw-r----- a 0 glenda glenda   0 Aug 16 12:06 /dev/aoe/0.0/data
  d-r-xr-xr-x a 0 glenda glenda   0 Aug 16 12:06 /dev/aoe/0.0/devlink
  --rw-r--r-- a 0 glenda glenda 512 Aug 16 12:06 /dev/aoe/0.0/ident



I can mail the old vblade binary to you if wanted.
-- 
"on travel, off the network ... and a fossil in my pocket"

Re: [9fans] aoe on sources

[erik quanstrom]

> i started vblade with:
>     cpu% dd -if /dev/zero -of 80.74 -bs 1024 -count 102400
>     102400+0 records in
>     102400+0 records out
>     cpu% vblade -ia 80.74 80.74
>     lblade 80.74 204672 sectors

looks good.

> mounted the target with:
>     term% bind -b '#ae' /dev

i think you mean '#æ'?

>     term% echo bind '#l/ether0' > /dev/aoe/ctl
>     term% echo config switch on spec z type aoe//dev/aoe/80.74 > /dev/sdctl
>     term% ls /dev/sdz0
>     /dev/sdz0/ctl
>     /dev/sdz0/raw
>     term% grep nopen /dev/aoe/80.74/ctl
>     nopen: 1

hmmm.  i couldn't reproduce this bug.  you can try to

	echo identify>/dev/aoe/80.74/ctl

and check the reported size of the target with

	ls -l /dev/aoe/80.74

the size of the data file will be the size of the target.

i did make some small changes to vblade.  if you continue to
have trouble, you might try again with the new vblade.  but
i don't think i fixed anything critical.

- erik

Re: [9fans] aoe on sources

[YAMANASHI Takeshi]

Hi,

I'm trying AoE but no success so far.  The sdata initiator succeeded to
connect to the AoE target presented by Plan 9 vblade,
but i didn't see /dev/sdl0/data while /dev/sdl0/^(ctl raw) were there.
What am i missing?


i started vblade with:
    cpu% dd -if /dev/zero -of 80.74 -bs 1024 -count 102400
    102400+0 records in
    102400+0 records out
    cpu% vblade -ia 80.74 80.74
    lblade 80.74 204672 sectors


mounted the target with:
    term% bind -b '#ae' /dev
    term% echo bind '#l/ether0' > /dev/aoe/ctl
    term% echo config switch on spec z type aoe//dev/aoe/80.74 > /dev/sdctl
    term% ls /dev/sdz0
    /dev/sdz0/ctl
    /dev/sdz0/raw
    term% grep nopen /dev/aoe/80.74/ctl
    nopen: 1

-- 
YAMANASHI Takeshi

Re: [9fans] aoe on sources

[erik quanstrom]

first, many thanks to geoff for spending quite a bit of time
making sure everything was right for inclusion in the sources tree.

> 
> There is at least one important fix in our version of the
> Coraid code, so if you're running old Coraid code, you should
> update.

fortunately, the bug in question never saw the light of day; nobody
is using it.  it was a mistake in sdaoe (not devaoe) decoding scsi16
commands which are only used on devices > 2TB.

- erik

[9fans] aoe on sources

[geoff]

I've incorporated Coraid's AoE code into our kernel.
Code and manual pages (aoe(3) and sdaoe(3)) are on sources now.
To use it, you need to include

dev
	# ...
	aoe
misc
	# ...
	sdaoe

in your kernel configuration(s), and follow the
instructions in the above-mentioned manual pages.

It's very lightly tested so far, but I'll exercise it against
a Coraid EtherDrive soon.

There is at least one important fix in our version of the
Coraid code, so if you're running old Coraid code, you should
update.