Hi Nathan, an Elasticsearch index consists of 1 or more primary shards and 0 or more replica shards. Documents are being assigned to shards according to their ID (or rather a hash of their ID) so that ideally the documents of one index are well-balanced across all shards (which can be hosted on different Elasticsearch nodes).
If you have one or more replica shards (basically just a copy of the primary shards), the absence of one or more nodes (depending on the total cluster size) can be compensated. See https://www.elastic.co/guide/en/elasticsearch/reference/2.3/_basic_concepts.html and https://www.elastic.co/guide/en/elasticsearch/guide/2.x/scale.html for more detailed information. Cheers, Jochen On Thursday, 28 July 2016 21:44:15 UTC+2, Nathan Mace wrote: > > OK, thanks! > > Semi-related question, when you have multiple ES nodes in a cluster how is > the load balanced? Is it like an even split across all the nodes (50/50 or > 25/25/25/25)? And how does it handle searches when one ES node is down? > Is the data on that node also saved on another node in the cluster for > redundancy? Sorry if this is not the place for these questions but this is > literally the only time I have ever dealt with Elastic Search. > > Thanks > > Nathan > > On Thursday, July 28, 2016 at 11:16:10 AM UTC-4, Jochen Schalanda wrote: >> >> Hi Nathan, >> >> the two configuration settings you've mentioned, >> elasticsearch_discovery_zen_ping_multicast_enabled and >> elasticsearch_discovery_zen_ping_unicast_hosts, are from the Graylog >> configuration file and don't need to be changed when adding another >> Elasticsearch node. >> >> Simply make sure, that the new Elasticsearch node has the same cluster >> name, can connect to all other Elasticsearch nodes (including Graylog >> itself), and has a list of the other Elasticsearch nodes to connect to. >> >> See >> http://docs.graylog.org/en/2.0/pages/configuration/elasticsearch.html#configuration >> >> for further details. >> >> Cheers, >> Jochen >> >> On Thursday, 28 July 2016 16:52:16 UTC+2, Nathan Mace wrote: >>> >>> I just installed Graylog + Mongo DB + Elastic Search on a Cent OS VM, >>> following the Graylog official documentation. Working great. >>> >>> However, when the time comes that it isn't feasible to add additional >>> CPU and RAM to the VM and I need to add another Elastic Search Node, what >>> is the procedure for doing that to an existing "cluster" (even if that >>> existing cluster is a single VM). >>> >>> I think it would be something along these lines: >>> >>> Create VM, install Elastic Search >>> Edit the Elastic Search config so it has the same cluster name as the >>> existing nodes >>> Enable the following options: >>> >>> elasticsearch_discovery_zen_ping_multicast_enabled = false >>> elasticsearch_discovery_zen_ping_unicast_hosts = 127.0.0.1:9300 >>> >>> >>> Start the new service >>> Restart the existing Graylog + Elastic Search services. >>> >>> Am I missing anything? >>> >>> Nathan >>> >> -- You received this message because you are subscribed to the Google Groups "Graylog Users" group. To unsubscribe from this group and stop receiving emails from it, send an email to [email protected]. To view this discussion on the web visit https://groups.google.com/d/msgid/graylog2/b3b48816-65a2-4a1b-b7ab-ad9d320bd7be%40googlegroups.com. For more options, visit https://groups.google.com/d/optout.
