Help with LaunchPad

[John Arbash Meinel]

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

SuperMMX wrote:
| Hi, John Arbash Meinel <john at arbash-meinel.com> :
|
| On Sat, 22 Mar 2008 09:11:14 -0500
| John Arbash Meinel <john at arbash-meinel.com> wrote:
|
|> Can you tell us which branch this is? I haven't seen this particular failure
|> before, and I would like to investigate.
|
| ~supermmx/jaccent/trunk-corrupt
|
|> Specifically, it shouldn't be possible to ^C a push and end up with something
|> corrupt. As if the push is interrupted we should just not reference the
|> incomplete data.
|
| Here is the steps (on Windows) :
|
| bps is the alias of "bzr push"
|
| 1) time bps --remember
bzr+ssh://supermmx@bazaar.launchpad.net/~supermmx/jaccent/trunk
| Unable to obtain lock lp--1217904948:///lock
| held by supermmx at gmail.com on host xuama-wxp02 [process #17416]
| locked 8 minutes, 48 seconds ago
| Will continue to try until 10:24:39
|

Well, you have a lock left over, so it shouldn't be able to push anything. But
after you did 'break-lock' and then pushed it should work.
...


| 4) bzr break-lock bzr+ssh://supermmx@bazaar.launchpad.net/~supermmx/jaccent/trunk
|
| Break lock
bzr+ssh://supermmx@bazaar.launchpad.net/%7Esupermmx/jaccent/trunk/.bzr/branch/lock
| held by supermmx at gmail.com on host xuama-wxp02 [process #17416]
| locked 25 minutes, 18 seconds ago? [y/n]: y
|
| 5) time bps --remember --overwrite
bzr+ssh://supermmx@bazaar.launchpad.net/~supermmx/jaccen
| t/trunk
| Pushed up to revision 3.
|
| real    3m9.047s
| user    0m0.031s
| sys     0m0.046s
|
| So, here only the last 3 revisions are pushed, while there should be 178.
|

So from here I see 178:
http://bazaar.launchpad.net/~supermmx/jaccent/trunk/changes


The only thing I can think of is that you encountered the delay from pushing to
bzr+ssh:// and when it gets mirrored to the http:// side.

Launchpad specifically has a policy of not directly accessing pull side from the
push side. (That way there is the opportunity to filter, and people can't just
upload any .jpg images and have them hosted.)

I believe the mirroring process is triggered at the end of the push, so it
should only be a minute or 2 from the time you push until it is accessible on
for anonymous http access.


I *do* see that the trunk-corrupt branch only has 3 revisions:
http://bazaar.launchpad.net/~supermmx/jaccent/trunk-corrupt/changes

I also see the failure that branching/checking out the trunk-corrupt branch
fails with a missing file-id.

Looking at the repository you are using the older Knit format (I would recommend
upgrading to --pack-0.92 if it is possible for you to do so.)

I certainly can say that the knit file for:
bzrignore-20070804125105-8q8tup8ey9xttuqo-1

Is not present. I don't quite understand how the push succeeded (at least to
revision 3), without actually pushing all of the content.

I can see that you have 181 full revisions in inventory.kndx, so somehow the
inventory managed to be completely copied, but it didn't manage to copy all of
the texts.

Andrew... could this be a bug in the get_data_stream() code? Is it actually
installing the inventories in Knit repositories rather than waiting until all
the texts are installed? I know it works correctly for pack repositories, but
that is an illegal ordering for Knit ones. (With the old copy code we would
buffer them in memory until the texts were copied, and then we would install the
inventories, which is what we need to preserve.)

John
=:->
-----BEGIN PGP SIGNATURE-----
Version: GnuPG v1.4.5 (Cygwin)
Comment: Using GnuPG with Mozilla - http://enigmail.mozdev.org

iD8DBQFH5SEzJdeBCYSNAAMRAlf8AJ4lJvNkxsyKf44U66Vr+Qb3OVz7eQCglCVd
v+nt2X7tx3ce3FYLsd0yDiE=
=Atuw
-----END PGP SIGNATURE-----