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
Currently, both the NativeUsersStore and NativeRolesStore can undergo
multiple state transitions. This is done primarily to check if the
security index is usable before it proceeds. However, such checks are
only needed for the tests, because if the security index is unavailable
when it is needed, the downstream actions invoked by the
NativeUsersStore and NativeRolesStore will throw the appropriate
exceptions notifying of that condition. In addition, both the
NativeUsersStore and NativeRolesStore had much duplicate code that
listened for cluster state changes and made the exact same state
transitions.
This commit removes the complicated state transitions in both classes
and enables both classes to use the SecurityTemplateService to monitor
all of the security index lifecycle changes they need to be aware of.
This commit also moves the logic for determining if the security index
needs template and/or mapping updates to the SecurityLifecycleService,
and makes the NativeRealmMigrator solely responsible for applying the
updates.
Original commit: elastic/x-pack-elasticsearch@b31d144597
Adds a new `xpack.security.authc.accept_default_password` setting that defaults to `true`. If it is set to false, then the default password is not accepted in the reserved realm.
Adds a bootstrap check that the above setting must be set to `false` if security is enabled.
Adds docs for the new setting and bootstrap.
Changed `/_enable` and `/_disable`, to store a blank password if the user record did not previously exist, which is interpreted to mean "treat this user as having the default password". The previous functionality would explicitly set the user's password to `changeme`, which would then prevent the new configuration setting from doing its job.
For any existing reserved users that had their password set to `changeme`, migrates them to the blank password (per above paragraph)
Closes: elastic/elasticsearch#4333
Original commit: elastic/x-pack-elasticsearch@db64564093
This commit adds basic tests that store a user and a role using the native API. The test checks
that the user and role can be used prior to starting the upgrade. The realm and roles caches are
also cleared to ensure the next authentication will require a read from the security index; this
ensures we are actually testing reads from the index.
Original commit: elastic/x-pack-elasticsearch@396862da94
When one of the 2 nodes in the old cluster is shut down, shards that were on that node will become unassigned and be marked to be
delay-allocated, i.e. either a node with shard data for that shard must be available or the allocation of the shards will be delayed for a minute.
In the mixed cluster the replica shard might not be allocated as the primary is already on the node with the newer version and replicas are not allowed
then to be allocated to a node of an older version of ES. Once both nodes are upgraded, the delay might still be in place, and can only be nullified if there
is shard data available on the node. If there never was a shard on that node though, it will take a minute and run into the timeout checking for green.
This commit ensures that all shards are fully-allocated before we do the rolling restart scenario
Original commit: elastic/x-pack-elasticsearch@a0d9b1b043
If the primary shard of an index with (number_of_replicas > 0) ends up on a new node in a mixed cluster, the replica cannot be allocated to the old node as
the new node might have written segments that use a new postings format or codec that is not available on the older node.
As x-pack automatically creates indices with number_of_replicas > 0, for example monitoring-data-*, the test can only wait for yellow in a mixed cluster.
Original commit: elastic/x-pack-elasticsearch@945d9e3811
This commit adds a timeout to the cluster health call that we wait on so that we can
see the status of the health request instead of getting timeouts failures with no
information to go on.
Original commit: elastic/x-pack-elasticsearch@2f34d01e00
This change allows reads of our native users and roles when the template version has not been updated to
match the current version. This is useful for rolling upgrades where the nodes are also being actively
queried and/or indexed into. Without this, we can wreak havoc on a cluster by causing exceptions during
replication, which leads to shard failures. On nodes that match the version defined in the template,
write operations are allowed since we know that we are backwards compatible in terms of format but we
may have added new fields and shouldn't index them until the mappings and template have been updated.
As part of this, the rolling upgrade tests from core were used as the basis for a very basic set of tests
for doing a rolling upgrade with x-pack.
Closeselastic/elasticsearch#4126
Original commit: elastic/x-pack-elasticsearch@9be518ef00
Alexander Reelsen
Ryan Ernst
Ali Beyad
javanna
Tim Vernum
Jason Tedor
Lee Hinman
Jay Modi
Nik Everett
Yannick Welsch
jaymode