The api jar was added for xpack extensions. However, extensions have
been removed in favor of using SPI, and the individual xpack jars like
core and security are published to enable this. This commit removes the
api jar, and switches the transport client to use the core jar (which
the api jar was just a rename of).
Original commit: elastic/x-pack-elasticsearch@58e069e66c
There were a number of leftover unnecessary elements in the module
build.gradle files that were holdovers from copying the original plugin
build.gradle. This commit removes these elements.
Original commit: elastic/x-pack-elasticsearch@08babbd520
For the idp-fixture (OpenLDAP + SAML), we have been generating the CA as part of the provisioning steps for the VM and then adding it to the test resources for the gradle project.
This meant that test-resources were dependent on vagrant provision, and as a consequence vagrant would download and provision the box during precommit. A bad thing (TM)
This change introduces a pre-generated CA, which is supplied to the VM instead so the tests only depend on fixed resources.
(The SAML integration test still uses the generated IdP Metadata file, but it copies it as part integ-test cluster setup, and doesn't treat it as a gradle "test resource")
Original commit: elastic/x-pack-elasticsearch@a352bf2a1f
When I dropped the CI script changes for SQL I forgot to restore the old
behavior. This fixes that.
Original commit: elastic/x-pack-elasticsearch@0e18b41464
The docs include portions of the SQL tests and for that to work they
need to point to position of the tests. They use a relative directory
but relative to *what*? That turns out to be a fairly complex thing to
answer, luckilly, `index.x.asciidoc` defines `xes-repo-dir` which points
to the root of the xpack docs. We can use that to find the sql tests
without having to answer the "relative to what?" question in two places.
Original commit: elastic/x-pack-elasticsearch@ebea586fdf
Calling start() when already in the STARTING state doesn't do anything, so the component
gets stuck in STARTING state forever.
Also: wait on the required index name not just the cluster.
Also: added more logging to help diagnose such issues (either in RemoteIndexAuditTrailStartingTests or production)
Original commit: elastic/x-pack-elasticsearch@fb81214fe7
The old home made sense before x-pack was split. The new home lines up
with where security keeps its scripts.
The jar file wasn't being included any more so this re-adds that.
Original commit: elastic/x-pack-elasticsearch@d3ec941397
This commit wraps the TestCluster in one that does not check stats to prevent tripping an
assertion for the accounting breaker as the index audit trail is running in the background and
will affect the value of the breakers.
See elastic/x-pack-elasticsearch#157
Original commit: elastic/x-pack-elasticsearch@381ebbd413
This commit adds special handling for null values when building a DistinguishedNamePredicate for
role mapping. Previously this would have resulted in an exception from the unboundid ldapsdk as
the DN is invalid.
relates elastic/x-pack-elasticsearch#3787
Original commit: elastic/x-pack-elasticsearch@9386dae03e
Includes:
- docs for new realm type "saml"
- docs for new settings for SAML realms
- a guide for setting up SAML accross ES + Kibana
Original commit: elastic/x-pack-elasticsearch@85f8f6d409
This allows any datetime function to be present in `EXTRACT` which feels
more consistent. `EXTRACT(FOO FROM bar)` is now just sugar for
`FOO(bar)`. This is *much* simpler to explain in the documentation then
"these 10 fields are supported by extract and they are the same as this
subset of the datetime functions."
The implementation of this is a little simpler then the old way. Instead
of resolving the function in the parser we create an
`UnresolvedFunction` that looks *almost* just like what we'd create for
a single argument function and resolve the function in the `Analyzer`.
This feels like a net positive as it allows us to group `EXTRACT`
resolution failures with other function resolution failures.
This also creates `UnresolvedFunctionTests` and
`UnresolvedAttributeTests`. I had to create `UnresolvedFunctionTests`
because `UnreolvedFunction` now has three boolean parameters which is
incompatible with the generic `NodeSubclassTests`'s requirement that all
ctor parameters be unique. I created `UnresolvedAttributeTests` because
I didn't want `UnresolvedFunctionTests` to call `NodeSubclassTests` and
figured that we'd want `UnresolvedAttributeTest` eventually and now felt
like as good a time as any.
Added a
Original commit: elastic/x-pack-elasticsearch@358aada308
We don't need the double quotes. Also, we follow up with an example that
shows how to write them in yml.
Original commit: elastic/x-pack-elasticsearch@835deca6f9
If any of the follow take place on security index, then any cached role mappings
are potentially invalid and the associated realms need to clear any cached users.
- Index recovers from red
- Index is deleted
- Index becomes out-of-date / not-out-of-date
Original commit: elastic/x-pack-elasticsearch@1bcd86fcd4
The HTTP Exporter in Monitoring allowed users to set a timeout parameters
for the requests. When set, this was setting the `master_timeout` query parameter
in Bulk Requests. The problem is that Bulk Requests do not support this type of
timeout.
Original commit: elastic/x-pack-elasticsearch@9be194006e
This adds support to allow different schemes in a proxy being used
compared to what the actual request requires. So if your proxy runs via
HTTP, but the endpoint you want to connect to uses HTTPS, this is now
possible to configure the proxy explicitely.
Also a small unit test for parsing this has been added.
relates elastic/x-pack-elasticsearch#3596
Original commit: elastic/x-pack-elasticsearch@176f7cdf0e
Tests have been failing because of out of order cluster state processing
or because of stopping/starting was still in progress. Current tests do
not do further stop/start tries after the first leading to potential
being stuck in an unwanted state. This commit removes the methods
checking for the state being started or stopped in favor of using the
stopWatcher/startWatcher methods which now check if the desired state is
reached and otherwise issue another start/stop command.
Original commit: elastic/x-pack-elasticsearch@97b3232a6a