Thank you Binoy. We are generating spellcheck source data: spellings_xxx.txt
by querying the main index only(we do have the field indexed in cloud). Due
to huge amount of data(160 million records), spellcheck build request taking
lot of time and consuming lot memory for index based spellcheck. So we have
query the filed value of one content type alone to build the
spellings_xxx.txt(1 million entries-14MB size).
Here is the behavior of the FileBasedSpellcheck on cloud:
Testing environment:
total : 4 solr instances : 4.10.4
External zookeeper ensemble: 3 instances
shard-1
-- leader1
-- replica1
shard2-2
-- leader2
-- replica2
we have pushed the file to cloud and sends the spellcheck build request to
leader-1 of shard-1.
1. First time, it has built the spellcheck index on leader-1 instance
shard-1 and replica-2 of shard2
2. Next time(cleaned the spellcheck index and restarted the cloud), have
noticed, it has built the index on leader-1 instance shard-1 and leader-2
instance shard-2(on both the leaders only)
Our spellcheck queries(posting to leader-1) are not returning any
suggestions , if we keep on refreshing the page.(may be when replicas get
the request)
Can you please let us know, is there way to built the filebased spellchekc
on cloud?.
Thanks
Riyaz
--
View this message in context:
http://lucene.472066.n3.nabble.com/FileBased-Spellcheck-on-Solr-cloud-tp4252034p4254432.html
Sent from the Solr - User mailing list archive at Nabble.com.
