> On Aug 24, 2014, at 1:13 AM, [email protected] wrote:
> 
> But the replicator won't hear about the document because it never changes. 
> And its not the first replication: the 300 documents were created on A and 
> replicated to B. 

How soon after pushing to B do you purge the documents? There's an 'echo' where 
the corresponding puller from B will see the documents that were just pushed to 
B. But that wouldn't cause a problem unless you purged the docs the instant you 
pushed them.

Since you mentioned 'peers' — if you've got a mesh of P2P replications, another 
scenario is that A pushes to B, then A purges the docs, but then meanwhile B 
pushes to C. If A later pulls from C it'll see and download the docs that it 
previously purged.

Otherwise, the only thing that would cause this would be if the replication 
checkpoint got lost somehow and the puller had to start from the beginning of 
the changes feed.

> I've solved the problem by adding a replication filter that skips old/paid 
> orders but I don't understand why that should be necessary.

That's the safest solution, I think. In general there's no guarantee that a 
purged doc will stay purged, due to the effects described above.

—Jens

-- 
You received this message because you are subscribed to the Google Groups 
"Couchbase Mobile" group.
To unsubscribe from this group and stop receiving emails from it, send an email 
to [email protected].
To view this discussion on the web visit 
https://groups.google.com/d/msgid/mobile-couchbase/9B83EFDF-B0D1-4312-948D-2C404D04357D%40couchbase.com.
For more options, visit https://groups.google.com/d/optout.

Reply via email to