[9fans] Replica (partly?) stuck in time

[9fans] Replica (partly?) stuck in time

[a]

On my cpu server, replica seems to be confused over its current
point in time.

When I first ran it today, I got a bunch of (correct) local
conflicts, and several files were updated. I resolved a few of
the local conflicts and re-ran, and while several things worked
fine, some odd behavior started showing.

Replica always tries to update three files, but never actually
does so. There were existing versions of the files; I've tried
removing them and recreating them. Replica notices the changes,
but still won't update them.

(pardon the bunch of -c's below. pull is a function calling
glenda's pull. this is as hostowner.)

cpu% pull -c acme/bin/386 -c 386/9pc -c 386/9load -c 386/bin -c 386/9loaddebug -c 386/9loadlite -c 386/9loadlitedebug -c 386/9pxeload -c 386/9pxeloaddebug -c rc/bin/9fs -c rc/bin/termrc -c rc/bin/cpurc -c rc/bin/cpurc.local -c lib/ndb/auth -c 386/9pc -c 386/9pc.gz -c 386/9pccpu -c 386/9pccpu.gz -c 386/9pcdisk -c 386/9pcdisk.gz -c 386/9pcf -c 386/9pcf.gz -c acme/mail/src
a sys/lib/sysconfig/auth/boundary/httppasswords 660 sys sys 1016833584
! sys/lib/sysconfig/auth/boundary/keys: locally created; will not overwrite
! sys/lib/sysconfig/auth/boundary/netkeys: locally created; will not overwrite

If I leave out the various -c arguments, replica has no idea
I've blessed those files. It's correct that httppasswords
doesn't exist while the other two do. If I give it a -s, It'll
make all three "a" lines; in any case, nothing in that
directory changes.

It is noticing other changes; for example, these two just
showed up in a successive run:

cpu% pull -c acme/bin/386 -c 386/9pc -c 386/9load -c 386/bin -c 386/9loaddebug -c 386/9loadlite -c 386/9loadlitedebug -c 386/9pxeload -c 386/9pxeloaddebug -c rc/bin/9fs -c rc/bin/termrc -c rc/bin/cpurc -c rc/bin/cpurc.local -c lib/ndb/auth -c 386/9pc -c 386/9pc.gz -c 386/9pccpu -c 386/9pccpu.gz -c 386/9pcdisk -c 386/9pcdisk.gz -c 386/9pcf -c 386/9pcf.gz -c acme/mail/src
a sys/lib/sysconfig/auth/boundary/httppasswords 660 sys sys 1016833584
! sys/lib/sysconfig/auth/boundary/keys: locally created; will not overwrite
! sys/lib/sysconfig/auth/boundary/netkeys: locally created; will not overwrite
c sys/man/2/atof
c sys/src/cmd/fossil/cache.c

Those files are indeed updated on my disk, and don't show up
in further replica runs. There are corresponding entries in
/dist/replica/client/plan9.(log db).

I can't find a definition of /dist/replica/client/plan9.time.
I noticed the permissions were weird for that directory (not
writable by hostowner), and the contents seems odd:
 1252443603        7182
The first number is in September of last year. I take the
next to be the "gen", although I'm not really clear how to
interpret that. Fixing permissions made no difference.

I'm stumped. Anyone have any ideas?

Re: [9fans] Replica (partly?) stuck in time

[ron minnich]

On Wed, May 19, 2010 at 10:08 PM,  <a@9srv.net> wrote:

> I'm stumped. Anyone have any ideas?


Yes, what I have done is stop using replica. I pull source from
bitbucket.org and build.

Replica is an interesting idea that does not work in the wide area. At
least not for me ...

ron

Re: [9fans] Replica (partly?) stuck in time

[David Leimbach]

[-- Attachment #1: Type: text/plain, Size: 857 bytes --]

On Wed, May 19, 2010 at 10:45 PM, ron minnich <rminnich@gmail.com> wrote:

> On Wed, May 19, 2010 at 10:08 PM,  <a@9srv.net> wrote:
>
> > I'm stumped. Anyone have any ideas?
>
>
> Yes, what I have done is stop using replica. I pull source from
> bitbucket.org and build.
>
> Replica is an interesting idea that does not work in the wide area. At
> least not for me ...
>
> ron
>
>
Who's updating the bitbucket.org sources, and how does one get started doing
that?

I've, after quite a long break, tried to get a Plan 9 CPU server up again in
a VM, and found the wiki instructions need some help, as they skip over a
few details needed to get a server up and running.

Maybe we need a "Guide" book.

Also I epically failed to use Maht's warlock for setting up a CPU server as
well, though I think it's a pretty great idea.

Dave

[-- Attachment #2: Type: text/html, Size: 1425 bytes --]

Re: [9fans] Replica (partly?) stuck in time

[David Leimbach]

[-- Attachment #1: Type: text/plain, Size: 1192 bytes --]

On Thu, May 20, 2010 at 8:23 AM, David Leimbach <leimy2k@gmail.com> wrote:

>
>
> On Wed, May 19, 2010 at 10:45 PM, ron minnich <rminnich@gmail.com> wrote:
>
>> On Wed, May 19, 2010 at 10:08 PM,  <a@9srv.net> wrote:
>>
>> > I'm stumped. Anyone have any ideas?
>>
>>
>> Yes, what I have done is stop using replica. I pull source from
>> bitbucket.org and build.
>>
>> Replica is an interesting idea that does not work in the wide area. At
>> least not for me ...
>>
>> ron
>>
>>
> Who's updating the bitbucket.org sources, and how does one get started
> doing that?
>
> I've, after quite a long break, tried to get a Plan 9 CPU server up again
> in a VM, and found the wiki instructions need some help, as they skip over a
> few details needed to get a server up and running.
>
> Maybe we need a "Guide" book.
>
> Also I epically failed to use Maht's warlock for setting up a CPU server as
> well, though I think it's a pretty great idea.
>
> Dave
>

How about a Google Wave for setting up Plan 9 CPU servers?  :-)

If you gave up on Wave previously, it's gotten a lot better in the last few
months...

It's also now open to anyone without invitation.

[-- Attachment #2: Type: text/html, Size: 2090 bytes --]

Re: [9fans] Replica (partly?) stuck in time

[ron minnich]

On Thu, May 20, 2010 at 3:23 PM, David Leimbach <leimy2k@gmail.com> wrote:

> Who's updating the bitbucket.org sources, and how does one get started doing
> that?

me.

I try to do it daily. There's still a few glitches (e.g. python on
plan 9 still has a way of exploding at times ... and 9vx can explode
too ...) so I have to do it by hand; I want it to be a cron job.

ron

Re: [9fans] Replica (partly?) stuck in time

[erik quanstrom]

> I can't find a definition of /dist/replica/client/plan9.time.
> I noticed the permissions were weird for that directory (not
> writable by hostowner), and the contents seems odd:
>  1252443603        7182
> The first number is in September of last year. I take the
> next to be the "gen", although I'm not really clear how to
> interpret that. Fixing permissions made no difference.
>
> I'm stumped. Anyone have any ideas?

it sounds like the fact that the plan9.time file doesn't
match the log is the problem.  it's not too hard to add
some prints to applylog to figure out why it is making
the decisions it makes.  perhaps there's already a verbose
option.

- erik