Hi, I am making some performance tests with a backup index from one week ago. For these tests I use a newly provisioned infrastructure identical to my production environment.
As my production collections have 2 shards each, I begin the test putting each backup shard in a different host so that I have scenario like: collection1 host1 : backup_shard1 host2 : backup_shard2 Each shard contains 200k documents and is 4GB large. Then, I issue a SPLITSHARD action at backup_shard1 with the command http://host1:8983/solr/admin/collections?action=SPLITSHARD&shard=backup_shard1&colection=collection1 The shard is split in two as expected, but none of the documents from the original shard go to the new ones even running a commit after the SPLITSHARD action. On another test, if I start both hosts with empty shards and begin to index documents, I get the right behaviour for SPLITSHARD, i.e., I can see the documents after the shards been split. So, what am I doing wrong at my backup recovery? How can I take shards from backup and split them on a different cluster than the original? PS: solr version is 4.9.0 -- Alexander Ramos Jardim
