Closed romanchyla closed 9 years ago
Do you mean that requests.post(SOLR_ENDPOINT)
within the /search
route should always speak to the same solr instance? Is this to take advantage of solr caching? Would the fact that internally, solr cloud picks a random shard to serve a request matter?
Does duration based stickiness make sense?
The request (if possible) should be routed to the same instance, at least for SOLR - that means that the api needs to pass appropriate headers, andthe load balancer at AWS can decide based on that. BTW: from what I see, the sessions are not sticky for AWS (which is good, IMO), but I didn't check solr