We’ve been running Solr Cloud with an Amazon load balancer (ALB rather than ELB) in prod for nearly a year. We send all updates through the load balancer without worrying about shards or leaders. It has been very reliable and very fast, loading about 1 million docs per minute. This is a 32 node cluster.
We run an ELB in the test environment, also very reliable. Test is an 8 node cluster. We have had one problem. For a full reload, 20 million docs, the commit was taking longer than the load balancer timeout, so the loader thought it failed, even though it succeeded. Using a TCP ELB sounds like a terrible idea. Use HTTP. wunder Walter Underwood wun...@wunderwood.org http://observer.wunderwood.org/ (my blog) > On Dec 26, 2017, at 1:05 PM, Rick Leir <rl...@leirtech.com> wrote: > > Per, > This is more of a question for the Drupal folks. But in passing, I would > suggest that you show your config and what you saw in your logs. And my guess > is firewall problems! > Cheers -- Rick > > On December 26, 2017 3:37:39 AM EST, Per Qvindesland <p...@me.com> wrote: >> Hi All >> >> I am trying to connect to a solrcloud through a elb from a Drupal 7 >> install, the elb is a tcp elb which seems to work well, drupal says it >> can talk to the solr install through the elb but when I try to index >> nothing seems to happen. >> >> Does anyone have any ideas on how to resolve this? or have any other >> suggestions on how to achieve redundancy? for the moment I am >> connecting to the solrcloud to one of the solrcloud instances, but >> should that fail then I don’t think I would have any redundancy. >> >> Regards >> Per > > -- > Sorry for being brief. Alternate email is rickleir at yahoo dot com
