[9fans] Pull?

[9fans] Pull?

[Kris Van Hees]

Is there potentially something down at the labs?  I've bring trying to execute
a pull for a while now tonight, and it is simply not authenticating me.  I also
created a new update account just to verify, and it won't accept it.

Anyone else experiencing problems?

	Kris

Re: [9fans] Pull?

[rob pike, esq.]

> Is there potentially something down at the labs?  I've bring trying to execute
> a pull for a while now tonight, and it is simply not authenticating me.  I also
> created a new update account just to verify, and it won't accept it.
>
> Anyone else experiencing problems?

i've seen this intermittently in the last few days.

-rob

Re: [9fans] Pull?

[mirtchov]

> Anyone else experiencing problems?

it worked for me just now, with a clean instance of factotum...

my problem is that I don't see any of the changes I've submitted
recently making it to sources (they're minor fixes to man pages
mostly, so there's no reason they should be immediately discarded :)

andrey

Re: [9fans] Pull?

[david presotto]

I'm out of town.  I'll stick all the changes on when I actually get 2 days
in a row near a
computer I understand.

----- Original Message -----
From: <mirtchov@cpsc.ucalgary.ca>
To: <9fans@cse.psu.edu>
Sent: Thursday, October 23, 2003 8:59 PM
Subject: Re: [9fans] Pull?


> > Anyone else experiencing problems?
>
> it worked for me just now, with a clean instance of factotum...
>
> my problem is that I don't see any of the changes I've submitted
> recently making it to sources (they're minor fixes to man pages
> mostly, so there's no reason they should be immediately discarded :)
>
> andrey
>
>
>

[9fans] Pull?

[Dan Cross]

Uh, I just ran pull and it deleted nearly every file on my system; something
tells me this is not normal.  Anyone else just have this?

	- Dan C.

Re: [9fans] Pull?

[Paul Lalonde]

-----BEGIN PGP SIGNED MESSAGE-----
Hash: SHA1

I just backed up my VM and gave it a shot.  I confirm that  
behaviour.  Yikes.

On 21-Dec-06, at 4:53 PM, Dan Cross wrote:

> Uh, I just ran pull and it deleted nearly every file on my system;  
> something
> tells me this is not normal.  Anyone else just have this?
>
> 	- Dan C.
>

-----BEGIN PGP SIGNATURE-----
Version: GnuPG v1.4.5 (Darwin)

iD8DBQFFiy9QpJeHo/Fbu1wRAiSAAKDQBta1uaJgXnaodTYISshTh65J9QCeNM+i
gDkK8IjWsyAR76GR+Pu/pUE=
=9yFR
-----END PGP SIGNATURE-----

Re: [9fans] Pull?

[Dan Cross]

On Thu, Dec 21, 2006 at 05:05:20PM -0800, Paul Lalonde wrote:
> I just backed up my VM and gave it a shot.  I confirm that  
> behaviour.  Yikes.

So, anyone else have any idea what's up with this?  Or, more importantly,
is sources repopulated?

	- Dan C.

Re: [9fans] Pull?

[Axel Belinfante]

> > I just backed up my VM and gave it a shot.  I confirm that  
> > behaviour.  Yikes.
> 
> So, anyone else have any idea what's up with this? 

/n/sources/plan9/dist/replica/plan9.log contains near the end
a whole series of d(elete) entries of which the following is the last

1166745605 15228 d 386/bin/ape/basename - 775 sys sys 1136656314 0

followed by a whole series of a(dd) entries of which following is the last

1166749207 15228 a dist/replica/plan9.proto - 664 sys sys 1144683345 2813


> Or, more importantly, is sources repopulated?

it looks like it.

Axel.

Re: [9fans] Pull?

[Aki Nyrhinen]

i just killed my system by doing a pull.

pull does not work.

On 12/22/06, Axel Belinfante <Axel.Belinfante@cs.utwente.nl> wrote:
> > > I just backed up my VM and gave it a shot.  I confirm that
> > > behaviour.  Yikes.
> >
> > So, anyone else have any idea what's up with this?
>
> /n/sources/plan9/dist/replica/plan9.log contains near the end
> a whole series of d(elete) entries of which the following is the last
>
> 1166745605 15228 d 386/bin/ape/basename - 775 sys sys 1136656314 0
>
> followed by a whole series of a(dd) entries of which following is the last
>
> 1166749207 15228 a dist/replica/plan9.proto - 664 sys sys 1144683345 2813
>
>
> > Or, more importantly, is sources repopulated?
>
> it looks like it.
>
> Axel.
>

Re: [9fans] Pull?

[Martin Neubauer]

* Aki Nyrhinen (anyrhine@gmail.com) wrote:
> i just killed my system by doing a pull.
> 
> pull does not work.

Did you interrupt the pull? I tried it and it looked suspiciously like a
problem that happened a few month ago - first about everything gets nuked
and then the whole thing gets repopulated. I recall that Geoff had quite
some work to get it fixed then.

The fact that this time it happened just before both the weekend and
christmas certainly isn't helping much.

Re: [9fans] Pull?

[geoff]

There was a problem locally wherein losing a file server connection
was making replica/scan think that all the files had vanished.  I've
taken measures to prevent it happening in future, and as far as we
know, pull is working again.  If in doubt, try `pull -n' first.

Re: [9fans] Pull?

[Scott Schwartz]

| There was a problem locally wherein losing a file server connection
| was making replica/scan think that all the files had vanished.  I've
| taken measures to prevent it happening in future, 

What sort of measures?  I mean, is there a general purpose strategy that
you used, or is this a special case?

Re: [9fans] Pull?

[Gabriel Diaz]

Hello

I've got the same problem too, pull -n shows that it want to remove  a
lot of stuff then, add a lot of files, should we let it do whatever it
wants or there is a way to re-sync the whole thing?

thanks

gabi


On 29 Dec 2006 00:11:24 -0500, Scott Schwartz <schwartz@bio.cse.psu.edu> wrote:
> | There was a problem locally wherein losing a file server connection
> | was making replica/scan think that all the files had vanished.  I've
> | taken measures to prevent it happening in future,
>
> What sort of measures?  I mean, is there a general purpose strategy that
> you used, or is this a special case?
>
>

Re: [9fans] Pull?

[geoff]

I did a pull last night to update a machine here and found that just
letting it run to completion was fine.  The deleted files that are
meant to persist get recreated.

I'm looking into the possibility of altering the replica log
to remove the deletions of files that are meant to persist.

Re: [9fans] Pull?

[Dan Cross]

On Fri, Dec 29, 2006 at 01:39:01PM -0500, geoff@plan9.bell-labs.com wrote:
> I did a pull last night to update a machine here and found that just
> letting it run to completion was fine.  The deleted files that are
> meant to persist get recreated.
> 
> I'm looking into the possibility of altering the replica log
> to remove the deletions of files that are meant to persist.

Unfortunately, it seems that I did a pull during a rather inconvenient
time when the files were deleted, but had not yet been re-added to the
replica log.  So, I fudged things around enough to get pull to sort of
work, but it was too late, and pull will not complete, and thus not
restore the system.  :-(

	- Dan C.

Re: [9fans] Pull?

[geoff]

I've removed the big block of deletions from
/n/sources/plan9/dist/replica/plan9.log, run replica/scan on sources
and pull -n on another machine, without any problems.

Try pull -n and if you're satisfied with its output, run pull.  If you
still get errors, mail them to me.

Re: [9fans] Pull?

[geoff]

[-- Attachment #1: Type: text/plain, Size: 262 bytes --]

It's somewhat specific to our situation, but the idea is that if you
`9fs server' and you know that there's always a specific file in the
server's root,

	if (test -e /n/server/known_file)
		foo /n/server

will execute foo only if the 9fs really worked.

[-- Attachment #2: Type: message/rfc822, Size: 2653 bytes --]

From: Scott Schwartz <schwartz@bio.cse.psu.edu>
To: Fans of the OS Plan 9 from Bell Labs <9fans@cse.psu.edu>
Subject: Re: [9fans] Pull?
Date: 29 Dec 2006 00:11:24 -0500
Date: Fri, 29 Dec 2006 00:11:24 -0500
Message-ID: <20061229051124.20307.qmail@g.galapagos.bx.psu.edu>

| There was a problem locally wherein losing a file server connection
| was making replica/scan think that all the files had vanished.  I've
| taken measures to prevent it happening in future, 

What sort of measures?  I mean, is there a general purpose strategy that
you used, or is this a special case?

Re: [9fans] Pull?

[Dan Cross]

On Thu, Dec 28, 2006 at 09:00:07PM -0500, geoff@plan9.bell-labs.com wrote:
> There was a problem locally wherein losing a file server connection
> was making replica/scan think that all the files had vanished.  I've
> taken measures to prevent it happening in future, and as far as we
> know, pull is working again.  If in doubt, try `pull -n' first.

I don't know; attempts to restore my system have so far been failures,
resulting in a host of error messages yet failing to repopulate the
filesystem.  :-(

	- Dan C.

Re: [9fans] Pull?

[Christian Grothaus]

> I don't know; attempts to restore my system have so far been failures,
> resulting in a host of error messages yet failing to repopulate the
> filesystem.  :-(

I have recovered my filesystem  by reformating the fossil writebuffer
with the contents of the last venti snapshot. (This may be a rather
drastic measure, though.)

Just for convenience, here's what I did:

(I have a standalone Plan 9 machine with a venti and fossil partition
on sdC0.)
Boot Plan 9 from a recent installation CD and enter the following
commands in a rio window:

# configure ip address and set venti environment variable
ip/ipconfig ether /net/ether0 192.168.1.2 255.255.255.0
venti='tcp!192.168.1.2!venti'

# save the venti config to a file, because venti needs it to start
venti/conf /dev/sdC0/arenas > /tmp/venti.conf

# then start venti
venti/venti -w -c /tmp/venti.conf -a $venti

# start fossil
fossil/fossil -f /dev/sdC0/fossil -c 'srv -p fscons'

# determine the score of the last archival snapshot
fossil/last /dev/sdC0/fossil
vac:a92cc4d2657044cfaadf3d4c9dbce3e89932c35a

# now format the writebuffer
fossil/flfmt -v a92cc...c35a /dev/sdC0/fossil

# log in the fossil console and halt fossil
con -l /srv/fscons
fsys main sync
fsys main halt

Cheers

Christian

Re: [9fans] Pull?

[Steve Simon]

> I have recovered my filesystem  by reformating the fossil writebuffer
> with the contents of the last venti snapshot. (This may be a rather
> drastic measure, though.)
> 
> Just for convenience, here's what I did:
> 

A bit late I know but in case it helps anyone else
I have written a hold-your-hand document on this after
doing it a couple of times myself:

http://plan9.bell-labs.com/sources/contrib/steve/doc/Venti-rescue.pdf

-Steve