Differential D17575 Diff 42283 src/docs/user/cluster/cluster.diviner

Changeset View

Standalone View

src/docs/user/cluster/cluster.diviner

	Show First 20 Lines • Show All 41 Lines • ▼ Show 20 Lines
	\| Service \| Setup \| Availability \| Loss Resistance \| Scalability			\| Service \| Setup \| Availability \| Loss Resistance \| Scalability
	\|---------\|-------\|--------------\|-----------\|------------			\|---------\|-------\|--------------\|-----------\|------------
	\| Databases \| Moderate \| High \| High \| Low			\| Databases \| Moderate \| High \| High \| Low
	\| Repositories \| Complex \| Moderate \| High \| Moderate			\| Repositories \| Complex \| Moderate \| High \| Moderate
	\| Daemons \| Minimal \| Low \| No Risk \| Low			\| Daemons \| Minimal \| Low \| No Risk \| Low
	\| SSH Servers \| Minimal \| Low \| No Risk \| Low			\| SSH Servers \| Minimal \| Low \| No Risk \| Low
	\| Web Servers \| Minimal \| High \| No Risk \| Moderate			\| Web Servers \| Minimal \| High \| No Risk \| Moderate
	\| Notifications \| Minimal \| Low \| No Risk \| Low			\| Notifications \| Minimal \| Low \| No Risk \| Low
				\| Fulltext Search \| Moderate \| High \| Minimal Risk \| Moderate
				epriestleyAuthorUnsubmitted Not Done Inline Actions I'd call this "No Risk" for loss resistance, since there's no data stored only in the search index (`bin/search init` + `bin/search index` can always completely rebuild it even if all the data from the fulltext index has been completely lost). (And maybe "Moderate" for availability, since Repositories only get a "Moderate" -- the other two "High" values make Phabricator 100% inaccessible if they go down.) epriestley: I'd call this "No Risk" for loss resistance, since there's no data stored only in the search…

	See below for a walkthrough of these services in greater detail.			See below for a walkthrough of these services in greater detail.


	Preparing for Clustering			Preparing for Clustering
	========================			========================

	To begin deploying Phabricator in cluster mode, set up `cluster.addresses`			To begin deploying Phabricator in cluster mode, set up `cluster.addresses`
	▲ Show 20 Lines • Show All 174 Lines • ▼ Show 20 Lines
	service provides no resistance to data loss because no data is at risk.			service provides no resistance to data loss because no data is at risk.

	Notification delivery normally requires very few resources, so adding more			Notification delivery normally requires very few resources, so adding more
	hosts is unlikely to have much impact on scalability.			hosts is unlikely to have much impact on scalability.

	For details, see @{article:Cluster: Notifications}.			For details, see @{article:Cluster: Notifications}.


				Cluster: Fulltext Search
				========================

				At a certain scale, you may begin to bump up against the limitations of MySQL's
				built-in fulltext search capabilities. We have seen this with very large
				installations with several million objects in the database and very many
				simultaneous requests. At this point you may consider adding Elasticsearch
				hosts to your cluster to reduce the load on your MySQL hosts.

				Elasticsearch has the ability to spread the load across multiple hosts and can
				handle very large indexes by sharding.

				Search does not involve any risk of data lost because it's always possible to
				rebuild the search index from the original database objects. This process can
				be very time consuming, however, especially when the database grows very large.

				With multiple Elasticsearch hosts, you can survive the loss of a single host
				with minimal disruption as Phabricator will detect the problem and direct
				queries to one of the remaining hosts.

				Phabricator supports writing to multiple indexing servers. This Simplifies
				Elasticsearch upgrades and makes it possible to recover more quickly from
				problems with the search index.

				For details, see @{article:Cluster: Search}.


	Overlaying Services			Overlaying Services
	===================			===================

	Although hosts can run a single dedicated service type, certain groups of			Although hosts can run a single dedicated service type, certain groups of
	services work well together. Phabricator clusters usually do not need to be			services work well together. Phabricator clusters usually do not need to be
	very large, so deploying a small number of hosts with multiple services is a			very large, so deploying a small number of hosts with multiple services is a
	good place to start.			good place to start.

	▲ Show 20 Lines • Show All 79 Lines • Show Last 20 Lines