Re: SOLR 4.4 - Slave always replicates full index

[Robin Woods <wo...@gmail.com>]

I did observe the same.. 

1. updated an existing document.. means potentially marking the previous
document as "deleted" and adding a new version of it.. posted the JSON doc
using the Documents interface on the Admin UI.. left the default commit
within "1000" ms there on the Documents UI..

2. NOT optimized

3. Slave seems to be triggering the full replication (full index seems to be
replicated)

4. Using Solr 4.9

5. Does have the following on the master (but commit within from the
Documents UI might be overloading this.. (mentioning here just in case)
<autoCommit> 
       <maxDocs>100000</maxDocs>
       <openSearcher>false</openSearcher> 
     </autoCommit>

Can someone look into this and suggest...

Thanks!




--
View this message in context: http://lucene.472066.n3.nabble.com/SOLR-4-4-Slave-always-replicates-full-index-tp4113089p4148526.html
Sent from the Solr - User mailing list archive at Nabble.com.

Re: SOLR 4.4 - Slave always replicates full index

[Robin Woods <wo...@gmail.com>]

Thanks Shawn. that makes sense.





--
View this message in context: http://lucene.472066.n3.nabble.com/SOLR-4-4-Slave-always-replicates-full-index-tp4113089p4148909.html
Sent from the Solr - User mailing list archive at Nabble.com.

Re: SOLR 4.4 - Slave always replicates full index

[Shawn Heisey <so...@elyograg.org>]

On 7/22/2014 5:00 PM, Robin Woods wrote:
> I think, I found the issue!
>
> I actually missed to mention a very important step that I did, which is,
> CORE SWAP
> otherwise, it's not replicating the full index.
>
> when we do CORE SWAP, doesn't it do the same checks of copying only deltas?

Yes, it will look for differences and only copy what's changed ... but
when you swap cores, you're pretty much guaranteed that the entire index
is different on the master compared to the slave, so it will have to
copy the entire thing.  Even if you build the index in exactly the same
way in two cores at exactly the same time on the same machine, the end
result will have minor differences, such as the timestamp on each file.

Thanks,
Shawn

Re: SOLR 4.4 - Slave always replicates full index

[Robin Woods <wo...@gmail.com>]

I think, I found the issue!

I actually missed to mention a very important step that I did, which is,
CORE SWAP
otherwise, it's not replicating the full index.

when we do CORE SWAP, doesn't it do the same checks of copying only deltas?





--
View this message in context: http://lucene.472066.n3.nabble.com/SOLR-4-4-Slave-always-replicates-full-index-tp4113089p4148678.html
Sent from the Solr - User mailing list archive at Nabble.com.