[tahoe-dev] Behavior when storage nodes go below shares.happy?

Zooko O'Whielacronx zooko at zooko.com
Fri Jul 9 16:29:21 UTC 2010


On Fri, Jul 9, 2010 at 9:50 AM, Kyle Markley <kyle at arbyte.us> wrote:
> tahoe version 1.7.0
> shares.needed = 2
> shares.happy = 4
> shares.total = 4
…
> during the night, leaving only 3 storage nodes running.

Uploads of immutable files should fail if happy = 4 and total servers
reachable = 3. Are you miscounting total servers reachable by
forgetting to count the node which is doing the upload, or something?

Is every immutable upload failing but the "backup job" is proceeding
to try the next and then the next immutable upload anyway?

I'm at a loss to otherwise explain this. Kevan and I made sure to
write super detailed unit tests proving that uploads will never appear
to succeed unless the servers-of-happiness criterion is met, so I
don't think that Tahoe-LAFS v1.7.0 uploader is making *that* mistake.
:-)

Regards,

Zooko



More information about the tahoe-dev mailing list