[tahoe-dev] odd repair output

[Greg Troxel]

With 1.9.2 (mac, pkgsrc, python 2.7), I am trying to maintain my bits on
the pubgrid.  Amazingly some of them have survived from the old
pubgrid.  (I know at least one server got moved from the old to the new,
with bits).

At the end of checking a directory, I see (with redaction of actual file
names):

'<root>': not healthy
 repair successful
'motd': not healthy
 repair successful
'motd/foo': healthy
'motd/bar: healthy
'motd/baz: healthy
'motd/bam: healthy
done: 6 objects checked
 pre-repair: 4 healthy, 2 unhealthy
 2 repairs attempted, 2 successful, 0 failed
 post-repair: 4 healthy, 2 unhealthy

I have in my .tahoe-config

shares.needed = 3
shares.happy = 6
shares.total = 10

and there are currently only 5 servers.  So I wonder if there is a "not
shares on 6" condition, leading to a repair, and that repair places
shares, and "succeeds" because it places them, but really it didn't
place them on enough servers.   So

  I wonder if the repair should not be attempted unless there are enough
  servers online to make the end-state possibly better than the
  beginning state.

  Probably the definition of "repair successful" should align better
  with post-repair being "healthy".

  I wonder how much of this will go away if extra shares are placed
  without version number changes in these kinds of situations.

  
-------------- next part --------------
A non-text attachment was scrubbed...
Name: not available
Type: application/pgp-signature
Size: 194 bytes
Desc: not available
URL: <http://tahoe-lafs.org/pipermail/tahoe-dev/attachments/20130316/3c08ceb0/attachment.pgp>