The .security index used several different types to differentiate the
documents added to the index (users, reserved-users, roles, etc). Since
types are deprecated in 6.x, this commit changes the .security index
access layer to only use a single type and have all documents in the
index be of that single type. To differentiate documents that may have
the same id (e.g. the same user name and role name), the appropriate
type of the document is prepended to the id. For example, a user named
"jdoe" will now have the document id "user-jdoe".
This commit also ensures that any native realm security index operations
that lead to auto creation of the security index first go through the process
of creating the internal security index (.security-v6) and creating the alias
.security to point to the internal index.
Lastly, anytime the security index is accessed without having been
upgraded, an exception is thrown notifying the user to use the
upgrade API to upgrade the security index.
Original commit: elastic/x-pack-elasticsearch@cc0a474aed
This commit adds new settings for the ssl keystore (not the ES keystore)
passphrase settings. New setting names are used, instead of trying to
support the existing names in both yml and the ES keystore, so that
there does not need to be complicated logic between the two. Note that
the old settings remain the only way to set the ssl passphrases for the
transport client, but the Settings object for transport clients are
created in memory by users, so they are already as "secure" as having a
loaded ES keystore. Also note that in the long term future (6.x
timeframe?) these settings should be deprecated and the keys/certs
themselves should be moved into the ES keystore, so there will be no
need for separate keystores/passphrases.
relates elastic/elasticsearch#22475
Original commit: elastic/x-pack-elasticsearch@be5275fa3d
The current testing setup only checked if watcher was started, but it
also needs to check for the index template in order to be sure that
everything is set up correctly, before trying to put a watch.
relates elastic/x-pack-elasticsearch#1762
Original commit: elastic/x-pack-elasticsearch@3ed78b15a1
Removes the `assemble` task from the `build` task when we have
removed `assemble` from the project. We removed `assemble` from
projects that aren't published so our releases will be faster. But
That broke CI because CI builds with `gradle precommit build` and,
it turns out, that `build` includes `check` and `assemble`. With
this change CI will only run `check` for projects without an
`assemble`.
Original commit: elastic/x-pack-elasticsearch@d01b0df1d9
Adds tests similar to `:qa:full-cluster-restart` for x-pack. You
run them with `gradle :x-pack:qa:full-cluster-restart:check`.
The actual tests are as basic as it gets: create a doc and load it,
shut down, upgrade to master, startup, and load it. Create a user
and load it, shut down, upgrade to master, startup, and load it.
Relates to elastic/x-pack-elasticsearch#1629
Original commit: elastic/x-pack-elasticsearch@8994bec8e7
This commit fixes the default password migration tests that had been failing reproducibly. The
first fix skips tests using the set enabled api when running against a version prior to 5.1.2 as
this api would otherwise trip an assertion that the xcontent builder was not closed. The second
fix is to ensure we set the password field in the user object.
relates elastic/x-pack-elasticsearch#1529
relates elastic/x-pack-elasticsearch#1516
Original commit: elastic/x-pack-elasticsearch@2f9c804309
In does not make sense for the time_field in the data_description to
be used as a by/over/partition field name, nor the summary_count_field,
categorization_field or as an influencer. Therefore, configurations
where the time_field in the data_description is used in the
analysis_config are now rejected.
Additionally, it causes a problem communicating with the C++ code if
the control field name (which is '.') is used in the analysis_config,
so this is also rejected at the validation stage.
Relates elastic/x-pack-elasticsearch#1684
Original commit: elastic/x-pack-elasticsearch@e6750a2cda
Removes the `assemble` task from projects that aren't published
to speed up `gradle assemble` so the unified release can call it.
Original commit: elastic/x-pack-elasticsearch@43dfcc15f3
This commit increases the amount of time to wait for green cluster health during a rolling upgrade
to account for the time that may be needed in the case of delayed shards. Additionally some old
timeout values were removed as they were used due to the default timeout of 30s.
Relates elastic/x-pack-elasticsearch#1683
Original commit: elastic/x-pack-elasticsearch@9996673db0
In 5.4.x, the datafeed attempts to get all fields from
doc_values by default. It has a `_source` parameter which
when enabled changes the strategy to instead try to get
all fields from the source.
This has been the most common issue users have been
reporting as it means the datafeed will fail to fetch
any text fields by default.
This change uses the field capabilities API in order
to automatically detect whether a field is aggregatable.
It then extracts such fields from doc_values while the
rest are taken from source. The change also adds
validation to the start datafeed action so that if
fields are missing mappings or the time field is not
aggregatable we respond with an appropriate error.
relates elastic/x-pack-elasticsearch#1649
Original commit: elastic/x-pack-elasticsearch@76e2cc6cb2
Real upgrades will have preserved their templates, so rolling upgrade tests should preserve them as well and internal services should be expected to replace them as needed.
Original commit: elastic/x-pack-elasticsearch@93a155951e
We wait for index templates to be installed before running tests, but
these can get upgraded when the master node is upgraded. If we don't
wait again in this case then tests can fail due to the cleanup code
being overwhelmed by pending cluster state updates that weren't waited
for before the test.
Original commit: elastic/x-pack-elasticsearch@eff0b24c11
Reworks the rolling restart tests so that all configuration
options share code. Now there is a project per configuration:
* `qa:rolling-upgrade:with-ssl-with-system-key`
* `qa:rolling-upgrade:with-ssl-without-system-key`
Original commit: elastic/x-pack-elasticsearch@5672b4a4f2
This commit adds better security for scroll requests in that they are now tied to a single user as
we only authorize the request that creates the scroll. This is accomplished by adding a
SearchOperationListener that listens for new scroll contexts and stores the authentication on the
ScrollContext. Then upon
retrieval of the search context for a query or fetch, the current authentication is compared to the
authentication that was present when the scroll context was created. If the current authentication
belongs to a different user, then a SearchContextMissingException will be thrown to prevent leaking
a valid vs invalid scroll id.
Additionally, signing of a scroll id is only performed when there is a older node in the cluster
that would expect the scroll id to be signed. Once this is backported to 5.x, we can remove this
bwc layer for 6.0/master.
Original commit: elastic/x-pack-elasticsearch@0e5dcafd32
The kibana user role in mixed cluster test expects a old cluster test to have run first that would
disable the kibana user, but that test is not executed on 5.1.1 and prior versions. This change
makes a update request an upsert to account for this.
relates elastic/x-pack-elasticsearch#1527
Original commit: elastic/x-pack-elasticsearch@df2945610e
Reworks the rolling restart tests so they'd have caugh an
incompatibility in the wire protocol that we say between 5.4.0 and
5.4.1.
Original commit: elastic/x-pack-elasticsearch@f5e69cf58e
This test failed, because there could have been other results in the
watch history. However by switching to the execute watch API, there
is no need to query the watch history at all. This also removes a 5
second wait time in the test.
relates elastic/x-pack-elasticsearch#1549
Original commit: elastic/x-pack-elasticsearch@692779521e
* Reduced a longish timeout to a shorter one, as a watch should be
executed in a HTTP test.
* Ensured that the TimeThrottleIntegration tests only query for own
watches in the watch history, also use random names for watch ids
* HipChatServiceTests configured deprecated logging package, so it was
not possible to follow the HTTP calls to the hipchat service endpoint.
relates elastic/x-pack-elasticsearch#1514
Relates elastic/x-pack-elasticsearch#1515
Original commit: elastic/x-pack-elasticsearch@adb492e4e9
The rolling upgrade nodes need a keystore for SSL configuration but
there was no dependency on the task that copies the keystore into the
output directory for the nodes to pick up as an extra configuration
file. This commit addresses this by adding such a dependency. To do
this, we need to break the dependency of the keystore copy task on the
REST spec copy task; this is not an issue since the dependency was for
convenience of ordering the task and not actually needed.
Original commit: elastic/x-pack-elasticsearch@fddbc06e9f
This commit marks the default password migration test as awaits fix as
the test is reliabily failling.
Original commit: elastic/x-pack-elasticsearch@4377480d98
The authentication object was changed in 5.4.0 in that it was conditionally signed depending on
the version and other factors. A bug was introduced however that causes the authentication to
actually get written with the version of the node it is being sent to even if that version is
greater than the version of the current node, which causes rolling upgrades to fail.
Original commit: elastic/x-pack-elasticsearch@a718ff8a52
This commit fixes the kibana write rolling upgrade test. The test needs to enable the kibana user
as another test explicitly disables it. After the test runs it disables the kibana user again.
relates elastic/x-pack-elasticsearch#1460
Original commit: elastic/x-pack-elasticsearch@437495432d
* Hide ML actions for tribe node client
* Remove unused parameters
* Enable ML actions and rest endpoints for the transport client
* Create the ML components for the transport client
* Add ml transport client tests
Original commit: elastic/x-pack-elasticsearch@509007ca29
Support the resolution of remote index names, including those that contain wildcards in the cluster name or index part)
Specifically these work:
- `GET /remote*:foo/_search`
- `GET /*:foo/_search`
- `GET /*:foo,*/_search`
- `GET /remote:*/_search`
- `GET /*:*/_search`
This change assumes that every user is allowed to attempt a cross-cluster search against any remote index, and the actual authorisation of indices happens on the remote nodes. Thus ` GET /*:foo/_search` will expand to search the `foo` index on every registered remote without consideration of the roles and privileges that the user has on the source cluster.
Original commit: elastic/x-pack-elasticsearch@b45041aaa3
As this does not require any reindexing this is easy to fix by just
changing the watch history template.
In addition the old templates are deleted on start up and the new ones
are instantiated.
Original commit: elastic/x-pack-elasticsearch@7e1ad495ad
If invalid job configs are transported to the master node then the root
cause of the validation exception gets reported as a remote_transport_exception,
which is extremely confusing.
This commit moves the validation of job configurations to the first node that
handles the action.
Fixeselastic/x-pack-kibana#1172
Original commit: elastic/x-pack-elasticsearch@5ed59d2a6f
* [ML] Not an error to close a job twice
* Error if job is opening
* Address review comments
* Test closed job isn’t resolved
Original commit: elastic/x-pack-elasticsearch@7da7b24c08
As fields with underscores will be disallowed in master, and we have to
prepare the upgrade, this commit renames the _status field to status.
When the 5.x upgrade logic is in place in the 5.x we can remove all the
old style _status handling from the master branch.
Note: All the BWC compatibility tests, that load 5.x indices are now
faking a finished upgrade by adding the `status` field to the mapping
of the watches index.
Original commit: elastic/x-pack-elasticsearch@9d5cc9aaec
The distribution of watches now happens on the node which holds the
watches index, instead of on the master node. This requires several
changes to the current implementation.
1. Running on shards and replicas
In order to run watches on the nodes with the watches index on its
primaries and replicas. To ensure that watches do not run twice, there is
a logic which checks the local shards, runs a murmurhash on the id and
runs modulo against the number of shards and replicas, this is the way to
find out, if a watch should run local. Reloading happens
2. Several master node actions moved to a HandledTransportAction, as they
are basically just aliases for indexing actions, among them the
put/delete/get watch actions, the acknowledgement action, the de/activate
actions
3. Stats action moved to a broadcast node action, because we potentially
have to query every node to get watcher statistics
4. Starting/Stopping watcher now is a master node action, which updates
the cluster state and then listeners acts on those. Because of this watches
can be running on two systems, if you those have different cluster state
versions, until the new watcher state is propagated
5. Watcher is started on all nodes now. With the exception of the ticker
schedule engine most classes do not need a lot of resources while running.
However they have to run, because of the execute watch API, which can hit
any node - it does not make sense to find the right shard for this watch
and only then execute (as this also has to work with a watch, that has not
been stored before)
6. By using a indexing operation listener, each storing of a watch now
parses the watch first and only stores on successful parsing
7. Execute watch API now uses the watcher threadpool for execution
8. Getting the number of watches for the stats now simply queries the
different execution engines, how many watches are scheduled, so this is
not doing a search anymore
There will be follow up commits on this one, mainly to ensure BWC compatibility.
Original commit: elastic/x-pack-elasticsearch@0adb46e658
Cross cluster search uses ClusterSearchShardsAction under the covers.
Without this change, you would need both "read_cross_cluster" and "view_index_metadata" privilegs in order to have permission to execute searches from a remote cluster.
Original commit: elastic/x-pack-elasticsearch@65a6aff329
This commit adds a token based access mechanism that is a subset of the OAuth 2.0 protocol. The
token mechanism takes the same values as a OAuth 2 standard (defined in RFC 6749 and RFC 6750),
but differs in that we use XContent for the body instead of form encoded values. Additionally, this
PR provides a mechanism for expiration of a token; this can be used to implement logout
functionality that prevents the token from being used again.
The actual tokens are encrypted using AES-GCM, which also provides authentication. The key for
encryption is derived from a salt value and a passphrase that is stored on each node in the
secure settings store. By default, the tokens have an expiration time of 20 minutes and is
configurable up to a maximum of one hour.
Relates elastic/x-pack-elasticsearch#8
Original commit: elastic/x-pack-elasticsearch@3d201ac2bf
This change removes the roles from the user in the remote cluster to show that the roles passed
from the source cluster are used in evaluating the user's permissions during multi-cluster search.
Original commit: elastic/x-pack-elasticsearch@35658168fb
We didn't realise it was possible for a qa module to depend on the
test classes of the plugin module, so we duplicated a test class.
But it turns out it IS possible to declare this dependency and avoid
the duplication.
Original commit: elastic/x-pack-elasticsearch@b6a21cda28
This commit removes the SecuredString class that was previously used throughout the security code
and replaces it with the SecureString class from core that was added as part of the new secure
settings infrastructure.
relates elastic/x-pack-elasticsearch#421
Original commit: elastic/x-pack-elasticsearch@e9cd117ca1
let close job and stop datafeed apis redirect to elected master node.
This is for cluster state observation purposes, so that a subsequent open and then close job or
start and then stop datafeed see the same local cluster state and sanity validation doesn't fail.
Original commit: elastic/x-pack-elasticsearch@21a63184b9
Ali Beyad
Ryan Ernst
Jason Tedor
Alexander Reelsen
Nik Everett
Martijn van Groningen
jaymode
David Roberts
Simon Willnauer
Jay Modi
James Baiera
Dimitris Athanasiou
Boaz Leskes
Chris Earle
David Kyle
Chris Earle
Jack Conradson
Tim Vernum
Tim Vernum
Yannick Welsch
Luca Cavanna