I'm also curious if this solution relies on a solr cloud on version n-1 sending
a distributed update request to a solr cloud on version n (or potnetially vice
versa) and how stable this is .. :-)
Sent from Bloomberg Professional for Android
- Original Message -
From: Gus Heck
To: users
That's interesting, but brings up the question of what happens if a node
(or the whole cluster) is rebooted in the middle of the process?
On Mon, Mar 31, 2025 at 10:02 PM Rahul Goswami
wrote:
> Some good points brought up in the discussion. The implementation we have
> reindexes a shard reading
This being an auxiliary process, for performance reasons, there is no
parallelism in the current version so as to take least resources away from
application processes. We upgrade core-by-core. The process reads a core
segment-by-segment and reindexes onto the same core. So the older version
of the