The introductory sections of the reference manual contains some simplified
instructions for adding a node to the cluster. Unfortunately they are a little
too simplified and only really work for clusters running on `localhost`. If you
try and follow these instructions for a distributed cluster then the new node
will, confusingly, auto-bootstrap itself into a distinct one-node cluster.
Multiple nodes running on localhost is a valid config, of course, but we should
spell out that these instructions are really only for experimentation and that
it takes a bit more work to add nodes to a distributed cluster. This commit
does so.
Also, the "important config" instructions for discovery say that you MUST set
`discovery.seed_hosts` whereas in fact it is fine to ignore this setting and
use a dynamic discovery mechanism instead. This commit weakens this statement
and links to the docs for dynamic discovery mechanisms.
Finally, this section is also overloaded with some technical details that are
not important for this context and are adequately covered elsewhere, and
completely fails to note that the default discovery port is 9300. This commit
addresses this.
Explicitly notes the Elasticsearch API endpoints that support CCS.
This should deter users from attempting to use CCS with other API
endpoints, such as `GET <index>/_doc/<_id>`.
Updates the cross-cluster search (CCS) documentation to note how
cluster-level settings are applied.
When `ccs_minimize_roundtrips` is `true`, each cluster applies its own
cluster-level settings to the request.
When `ccs_minimize_roundtrips` is `false`, cluster-level settings for
the local cluster is used. This includes shard limit settings, such as
`action.search.shard_count.limit`, `pre_filter_shard_size`, and
`max_concurrent_shard_requests`. If these limits are set too low, the
request could be rejected.
Today we use `cluster.join.timeout` to prevent nodes from waiting indefinitely
if joining a faulty master that is too slow to respond, and
`cluster.publish.timeout` to allow a faulty master to detect that it is unable
to publish its cluster state updates in a timely fashion. If these timeouts
occur then the node restarts the discovery process in an attempt to find a
healthier master.
In the special case of `discovery.type: single-node` there is no point in
looking for another healthier master since the single node in the cluster is
all we've got. This commit suppresses these timeouts and instead lets the node
wait for joins and publications to succeed no matter how long this might take.
Synced flush was a brilliant idea. It supports instant recoveries with a
quite small implementation. However, with the presence of sequence
numbers and retention leases, it is no longer needed. This change
removes it from 8.0.
Relates #5077
* Creates a prerequisites section in the cross-cluster replication (CCR)
overview.
* Adds concise definitions for local and remote cluster in a CCR context.
* Documents that the ES version of the local cluster must be the same
or a newer compatible version as the remote cluster.
The "Restore any snapshots as required" step is a trap: it's somewhere between
tricky and impossible to restore multiple clusters into a single one.
Also add a note about configuring discovery during a rolling upgrade to
proscribe any rare cases where you might accidentally autobootstrap during the
upgrade.
The docs specify that cluster.remote.connect disables cross-cluster
search. This is correct, but not fully accurate as it disables any
functionality that relies on remote cluster connections: cross-cluster
search, remote data feeds, and cross-cluster replication. This commit
updates the docs to reflect this.
Today the docs say that the low watermark has no effect on any shards that have
never been allocated, but this is confusing. Here "shard" means "replication
group" not "shard copy" but this conflicts with the "never been allocated"
qualifier since one allocates shard copies and not replication groups.
This commit removes the misleading words. A newly-created replication group
remains newly-created until one of its copies is assigned, which might be quite
some time later, but it seems better to leave this implicit.
Clarifies not to set `cluster.initial_master_nodes` on nodes that are joining
an existing cluster.
Co-Authored-By: James Rodewig <james.rodewig@elastic.co>
Setting `cluster.routing.allocation.disk.include_relocations` to `false` is a
bad idea since it will lead to the kinds of overshoot that were otherwise fixed
in #46079. This setting was deprecated in #47443. This commit removes it.
Setting `cluster.routing.allocation.disk.include_relocations` to `false` is a
bad idea since it will lead to the kinds of overshoot that were otherwise fixed
in #46079. This commit deprecates this setting so it can be removed in the next
major release.
We added docs for proxy mode in #40281 but on reflection we should not be
documenting this setting since it does not play well with all proxies and we
can't recommend its use. This commit removes those docs and expands its Javadoc
instead.
The changes in #32006 mean that the discovery process can no longer use
master-ineligible nodes as a stepping-stone between master-eligible nodes.
This was normally an indication of a strange and possibly-fragile configuration
and was not recommended. This commit clarifies that only master-eligible nodes
are now involved with discovery.
With this commit, Elasticsearch will no longer prefer using shards in the same location
(with the same awareness attribute values) to process `_search` and `_get` requests.
Instead, adaptive replica selection (the default since 7.0) should route requests more efficiently
using the service time of prior inter-node communications. Clusters with big latencies between
nodes should switch to cross cluster replication to isolate nodes within the same zone.
Note that this change only targets 8.0 since it is considered as breaking. However a follow up
pr should add an option to activate this behavior in 7.x in order to allow users to opt-in early.
Closes#43453
* Snapshot cleanup functionality via transport/REST endpoint.
* Added all the infrastructure for this with the HLRC and node client
* Made use of it in tests and resolved relevant TODO
* Added new `Custom` CS element that tracks the cleanup logic.
Kept it similar to the delete and in progress classes and gave it
some (for now) redundant way of handling multiple cleanups but only allow one
* Use the exact same mechanism used by deletes to have the combination
of CS entry and increment in repository state ID provide some
concurrency safety (the initial approach of just an entry in the CS
was not enough, we must increment the repository state ID to be safe
against concurrent modifications, otherwise we run the risk of "cleaning up"
blobs that just got created without noticing)
* Isolated the logic to the transport action class as much as I could.
It's not ideal, but we don't need to keep any state and do the same
for other repository operations
(like getting the detailed snapshot shard status)
Removes support for using a system property to disable the automatic release of
the write block applied when a node exceeds the flood-stage watermark.
Relates #42559
If a node exceeds the flood-stage disk watermark then we add a block to all of
its indices to prevent further writes as a last-ditch attempt to prevent the
node completely exhausting its disk space. However today this block remains in
place until manually removed, and this block is a source of confusion for users
who current have ample disk space and did not even realise they nearly ran out
at some point in the past.
This commit changes our behaviour to automatically remove this block when a
node drops below the high watermark again. The expectation is that the high
watermark is some distance below the flood-stage watermark and therefore the
disk space problem is truly resolved.
Fixes#39334
Uses JDK 11's per-socket configuration of TCP keepalive (supported on Linux and Mac), see
https://bugs.openjdk.java.net/browse/JDK-8194298, and exposes these as transport settings.
By default, these options are disabled for now (i.e. fall-back to OS behavior), but we would like
to explore whether we can enable them by default, in particular to force keepalive configurations
that are better tuned for running ES.
Today the lag detector may remove nodes from the cluster if they fail to apply
a cluster state within a reasonable timeframe, but it is rather unclear from
the default logging that this has occurred and there is very little extra
information beyond the fact that the removed node was lagging. Moreover the
only forewarning that the lag detector might be invoked is a message indicating
that cluster state publication took unreasonably long, which does not contain
enough information to investigate the problem further.
This commit adds a good deal more detail to make the issues of slow nodes more
prominent:
- after 10 seconds (by default) we log an INFO message indicating that a
publication is still waiting for responses from some nodes, including the
identities of the problematic nodes.
- when the publication times out after 30 seconds (by default) we log a WARN
message identifying the nodes that are still pending.
- the lag detector logs a more detailed warning when a fatally-lagging node is
detected.
- if applying a cluster state takes too long then the cluster applier service
logs a breakdown of all the tasks it ran as part of that process.
Most of the circuit breaker settings are dynamically configurable.
However, `indices.breaker.total.use_real_memory` is not. With this
commit we add a clarifying note that this specific setting is static.
Closes#44974
