Add back UpgradeReallyOldIndexTest from 1.x, but test 0.90.6 index
(Lucene 4.x) instead of 0.20 (Lucene 3.x), and make sure
only_ancient_segments works.
Closes#10639
This option defaults to false, because it is also important to upgrade
the "merely old" segments since many Lucene improvements happen within
minor releases.
But you can pass true to do the minimal work necessary to upgrade to
the next major Elasticsearch release.
The HTTP GET upgrade request now also breaks out how many bytes of
ancient segments need upgrading.
Closes#10213Closes#10540
Conflicts:
dev-tools/create_bwc_index.py
rest-api-spec/api/indices.upgrade.json
src/main/java/org/elasticsearch/action/admin/indices/optimize/OptimizeRequest.java
src/main/java/org/elasticsearch/action/admin/indices/optimize/ShardOptimizeRequest.java
src/main/java/org/elasticsearch/action/admin/indices/optimize/TransportOptimizeAction.java
src/main/java/org/elasticsearch/index/engine/InternalEngine.java
src/test/java/org/elasticsearch/bwcompat/StaticIndexBackwardCompatibilityTest.java
src/test/java/org/elasticsearch/index/engine/InternalEngineTests.java
src/test/java/org/elasticsearch/rest/action/admin/indices/upgrade/UpgradeReallyOldIndexTest.java
We have two completely different code paths for mappings updates, depending on
whether they come from the API or are guessed based on the parsed documents.
This commit makes dynamic mappings updates execute like updates from the API.
The only change in behaviour is that a document that fails parsing can not
modify mappings anymore (useful to prevent issues such as #9851). Other than
that, this change should be fairly transparent to users but working this way
opens doors to other changes such as validating dynamic mappings updates on the
master node (#8688).
The way it works internally is that Mapper.parse now returns a Mapper instead
of being void. The returned Mapper represents a mapping update that has been
performed in order to parse the document. Mappings updates are propagated
recursively back to the root mapper, and once parsing is finished, we check
that the mappings update can be applied, and either fail the parsing if the
update cannot be merged (eg. because of a concurrent mapping update from the
API) or merge the update into the mappings.
However not all mappings updates can be applied recursively, `copy_to` for
instance can add mappings at totally different places in the tree. Because of
it I added ParseContext.rootMapperUpdates which `copy_to` fills when the
field to copy data to does not exist in the mappings yet. These mappings
updates are merged from the ones generated by regular parsing.
One particular mapping update was the `auto_boost` setting on the `all` root
mapper. Being tricky to work on, I removed it in favour of search-time checks
that payloads have been indexed.
One interesting side-effect of the change is that concurrency on ObjectMapper
is greatly simplified since we do not have to care anymore about having
concurrent dynamic mappings and API updates.
Also added a couple nocommits for some issues with tests after mockfs is
working again. But I also re-enabled the mockfs suppression in the base
test case for now.
Allowing tests writing to the working directory can mask problems.
For example, multiple tests running in the same jvm, and using the
same relative path, may cause issues if the first test to run
leaves data in the directory, and the second test does not remember
to cleanup the path before using it.
This change adds security manager rules to disallow tests writing
to the working directory. Instead, tests create a temp dir with
the existing test framework.
closes#10605
This adds a new feature to the Term Vectors API which allows for filtering of
terms based on their tf-idf scores. With `dfs` option on, this could be useful
for finding out a good characteric vector of a document or a set of documents.
The parameters are similar to the ones used in the MLT Query.
Closes#9561
Using ThreadLocalRandom only prevents reproducibilty but doesn't buy us
anything. In production different datapaths won't have the same since
anyway or at least with a low likelyhood.
We need to preserve settings (yet transient) even though the engine is not yet
started. This commit moves back to a single EngineConfig to simplify IndexShard
and settings state.
Closes#10584
Local execution of transport messages failures can create a more detailed remote transport exceptions. Also, when failing to handle an exception, the error should be logged, and not call the handler again with another exception
closes#10554
This commit adds a `rewrite` parameter to the validate API in order to shown
how the given query is re-written into primitive queries. For example, an MLT
query is re-written into a disjunction of the selected terms. Other use cases
include `fuzzy`, `common_terms`, or `match` query especially with a
`cutoff_frequency` parameter. Note that the explanation is only given for a
single randomly chosen shard only, so the output may vary from one shard to
another.
Relates #1412Closes#10147
Today the engine writes the transaction log itself as well as manages
all the commit / translog mapping internally. Yet, if an engine is closed
and reopend it doesn't replay it's translog or does anything to be consistent
with it's latest state again.
This change moves the transaction log replay code into the Engine / InternalEngine
and adds unittests for replaying and consistency.
Closes#10452
At the moment, we are very strict when handling data folders containing corrupted shards and will fail any recovery attempt into it. Typically this wouldn't be a problem as the shard will be assigned to another node (which we try first anyway when a shard fails). However, it has been proven to be too strict for smaller clusters which may not have an extra node available (either because of allocation filtering, disk space issues etc.). This commit changes the behavior to force a full recovery. Once all the new files are verified we remove the old corrupted data and start the shard.
This also fixes a small issue where the shard state file wasn't deleted on an engine failure (we had a protection against deleting the state file on an active shard, but in this case the shard is still active but will be removed). The state deletion is also moved to before the failure handlers are called, to avoid race conditions when calling the master (it will potentially try to read it when allocating the shard)
Closes#10558
This changes adds the ability to specify the units for the x-axis for derivative values and calculate the derivative based on those units rather than the original histograms x-axis units
ShapeBuilder's coordinate parser expected 2 double values for every coordinate array. If > 2 doubles were provided the parser terminated parsing of the coordinate array. This resulted in an invalid Shape state leaving LineStrings, LinearRings, and Polygons with a single coordinate. An incorrect parse exception was thrown. This corrects the parser to ignore those values in the 3rd+ dimension, correctly parsing the rest of the coordinate array.
Unit tests have been updated to verify the fix.
closes#10510
Prevents the user from changing strategies, tree, tree_level or precision. distance_error_pct changes are allowed as they do not compromise the integrity of the index. A separate issue is open for allowing users to change tree_level or precision.
OGC SFA 2.1.10 assertion 3 allows interior boundaries to touch exterior boundaries provided they intersect at a single point. Issue #9511 provides an example where a valid shape is incorrectly interpreted as invalid (a false violation of assertion 3). When the intersecting point appears as the first and last coordinate of the interior boundary in a polygon, the ShapeBuilder incorrectly counted this as multiple intersecting vertices. The fix required a little more than just a logic check. Passing the duplicate vertices resulted in a connected component in the edge graph causing an invalid self crossing polygon. This required additional logic to the edge assignment in order to correctly segment the connected components. Finally, an additional hole validation has been added along with proper unit tests for testing valid and invalid conditions (including dateline crossing polys).
closes#9511
This is really a Collector instead of a filter. This commit deprecates the
`limit` filter, makes it a no-op and recommends to use the `terminate_after`
parameter instead that we introduced in the meantime.
When checking the JVM currently running ES we provide the user with
help on which environment variable to use to disable the check in
case the check fails. The variable we point to however is the wrong
one.
Most tests don't "really" need to fsync, and this is costly (makes
tests slower, wears out our SSDs).
This change makes it uncommon to actually fsync when Lucene asks for
it. It's just a workaround (in MockDirectoryHelper) until we can
cutover Elasticseach to use MockFileSystem like Lucene.
Closes#10516
This change logs total space, free space, usable free space, an
estimate of whether the IO system spins (e.g., SSD or not), the mount
point and filesystem type, on node startup.
It produces log output like this:
[2015-04-09 12:09:30,244][INFO ][env ] [node_t0] node data locations details:
-> /l/es.logspins/target/J0/data/TEST-haswell-CHILD_VM=[0]-CLUSTER_SEED=[2926863498862121027]-HASH=[AFC194B1B384B]/nodes/0, free_space [260.6gb], usable_space [256.3gb], total_space [465gb], spins? [no], mount [/ (/dev/mapper/haswell--vg-root)], type [btrfs]
Closes#10502
Today we force a flush before check index to ensure we have an index
to check on. Yet if the index is large and the FS is slow this can have
significant impact on the index deletion performance. This commit introduces
a check if there are any uncommitted changes in order to skip the additional commit.
Closes#10505
ReduceContext contains the list of aggregations to reduce but these aggregations are set as null half of the time. This change makes the reduce(ReduceContext) method changed to reduce(List<InternalAggregation>, ReduceContext) and ReduceContext now only holds the BigArrays and Script services.
* In code, we mark `River`, `AbstractRiverComponent`, `RiverComponent` and `RiverName` classes as deprecated
* We log that information when a cluster is still using it
* We add this information in the plugins list as well
Today we check every regular expression eagerly against every possible term.
This can be very slow if you have lots of unique terms, and even the bottleneck
if your query is selective.
This commit switches to Lucene regular expressions instead of Java (not exactly
the same syntax yet most existing regular expressions should keep working) and
uses the same logic as RegExpQuery to intersect the regular expression with the
terms dictionary. I wrote a quick benchmark (in the PR) to make sure it made
things faster and the same request that took 750ms on master now takes 74ms with
this change.
Close#7526
The refactoring in #9544 introduced a regression that broke multi-level
aggregations using breadth-first. This was due to sub-aggregators creating
deferred collectors before their parent aggregator and then the parent
aggregator trying to collect sub aggregators directly instead of going through
the deferred wrapper.
This commit fixes the issue but we should try to simplify all the pre/post
collection logic that we have.
Also `breadth_first` is now automatically ignored if the sub aggregators need
scores (just like we ignore `execution_mode` when the value does not make sense
like using ordinals on a script).
Close#9823
To ensure subclasses like MockInternalEngine which is in a different
package (test.engine) are logging under the same logger name this commit
moves to a static logger class to determin the logger name. This way
all subclasses of engine will log under `index.engine` which also plays
nicely with `@TestLogging` where log messages sometimes disappeared since
they were enabled for the `index.engine` package but not for `test.engine`
If the translog is buffered we must make sure everything is synced to disk
before we rollback the writer otherwise we open a window for potential dataloss due
to stupid errors preventing the translog from being closed.
Allows the user to calculate a Moving Average over a histogram of buckets. Provides four different
moving averages:
- Simple
- Linear weighted
- Single Exponentially weighted (aka EWMA)
- Double Exponentially weighted (aka Holt-winters)
Closes#10024
For bacwards compatibility reasons routing_nodes were previously printed out when routing_table was requested, together with the actual routing_table. Now they are printed out only when requests through `routing_nodes` flag.
Relates to #10412Closes#10486
Cluster state api returns both routing_table and routing_nodes sections whenever routing_table is requested. That is pretty much the same info, just grouped differently. This commit allows to differentiate between the two. Yet, routing_table still returns both for bw comp reasons.
Closes#10352Closes#10412
Removed the following methods from `ScriptService`, which don't require the `ScriptContext` argument:
```
public CompiledScript compile(String lang, String script, ScriptType scriptType)
public ExecutableScript executable(String lang, String script, ScriptType scriptType, Map<String, Object> vars)
public SearchScript search(SearchLookup lookup, String lang, String script, ScriptType scriptType, @Nullable Map<String, Object> vars)
```
Also removed the ScriptContext.Standard.GENERIC_PLUGIN enum value, as it was used only for backwards compatibility.
Plugins that make use of scripts should declare their own script contexts through `ScriptModule#registerScriptContext` and use them when compiling/executing scripts.
Closes#10476
Plugins can now define multiple operations/contexts that they use scripts for. Fine-grained settings can then be used to enable/disable scripts based on each single registered context.
Also added a new generic category called `plugin`, which will be used as a default when the context is not specified. This allows us to restore backwards compatibility for plugins on `ScriptService` by restoring the old methods that don't require the script context and making them internally use the `plugin` context, as they can only be called from plugins.
Closes#10347Closes#10419
Align get indexed scripts and get search template apis to our get api, which returns a response body when the document is not found, with a found boolean flag. Also, return metadata info all the time too.
Closes#7325Closes#10396
This tests adds a mappings with {"fielddata": {"format": "doc_values"}} but the
default mapping has {"doc_values": false} so when the document mapper parsing
logic merges both we have {"doc_values": false,"fielddata": {"format": "doc_values"}}
and {"doc_values": false} wins, so the test is not using doc values while it
thought it would.
In several places in the code we need to notify a node it needs to do something (typically the master). When that node is the local node, we have an optimization in serveral places that runs the execution code immediately instead of sending the request through the wire to itself. This is a shame as we need to implement the same pattern again and again. On top of that we may forget (see note bellow) to do so and we might have to write some craft if the code need to run under another thread pool.
This commit folds the optimization in the TrasnportService, shortcutting wire serliazition if the target node is local.
Note: this was discovered by #10247 which tries to import a dangling index quickly after the cluster forms. When sending an import dangling request to master, the code didn't take into account that fact that the local node may master. If this happens quickly enough, one would get a NodeNotConnected exception causing the dangling indices not to be imported. This will succeed after 10s where InternalClusterService.ReconnectToNodes runs and actively connects the local node to itself (which is not needed), potentially after another cluster state update.
Closes#10350
The exceptionCaught method had default access, which imposes a requirement
for subclasses that need to override this method to be in a specific package. This
change simply makes the method protected, which removes the package requirement.
We still have a lot of APIs that use setNextReader in order to change the
current segment that should be considered. This commit moves such APIs to
getLeafXXX() instead to be more in-line with Lucene 5's collector API.
I also renamed setDocId to setDocument to be more in-line with the doc values
APIs.
Close#10389
These tests create artificial hash collisions in order to make sure that they
can be resolved correctly. But this also makes the tests very slow if there
are too many collisions because insertions/deletions become linear in such
cases. The tests have been modified to not do too many iterations when
collisions are likely.
Close#10442
Closes#10435.
Squashed commit of the following:
commit aa1935c790b2731fc2bbc7de6142b09e3fe8bd4a
Author: Ryan Ernst <ryan@iernst.net>
Date: Mon Apr 6 13:44:40 2015 -0700
fix index lookup
commit bb6373595ff62ffc56fdf0cba3ac9c0ebe679946
Merge: 916962b eb3a170
Author: Robert Muir <rmuir@apache.org>
Date: Mon Apr 6 14:24:38 2015 -0400
Merge branch 'lucene_r1671277' of github.com:elasticsearch/elasticsearch into lucene_r1671277
commit 916962b82d192a53add471b4cc4a1396bc30eb0e
Merge: 197b3a2 21f72fe
Author: Robert Muir <rmuir@apache.org>
Date: Mon Apr 6 07:09:41 2015 -0400
Merge branch 'master' into lucene_r1671277
commit eb3a1703f7932ddd0cf3e83bec0e86131d255407
Author: Ryan Ernst <ryan@iernst.net>
Date: Sat Apr 4 11:06:03 2015 -0700
re-enable index lookup tests
commit 80d65d5eab39062dd8364687da74ddbb87ebcb76
Author: Ryan Ernst <ryan@iernst.net>
Date: Sat Apr 4 10:39:52 2015 -0700
update pom to point to new snapshot repo
commit 197b3a21ac2c2d70c9f740fe53e58632a22d1aad
Author: Robert Muir <rmuir@apache.org>
Date: Sat Apr 4 12:51:22 2015 -0400
fix postingsenum usage
commit 0e2b7a00cd07d068f755c51185ac521aa1eb0326
Author: Robert Muir <rmuir@apache.org>
Date: Sat Apr 4 12:21:23 2015 -0400
upgrade to lucene r1671277 (have not yet run tests or looked at postings changes)
The current implementation of AbstractBlobContainer.deleteByPrefix() calls AbstractBlobContainer.deleteBlobsByFilter() which calls BlobContainer.listBlobs() for deleting files, resulting in loading all files in order to delete few of them. This can be improved by calling BlobContainer.listBlobsByPrefix() directly.
This problem happened in #10344 when the repository verification process tries to delete a blob prefixed by "tests-" to ensure that the repository is accessible for the node. When doing so we have the following calling graph: BlobStoreRepository.endVerification() -> BlobContainer.deleteByPrefix() -> AbstractBlobContainer.deleteByPrefix() -> AbstractBlobContainer.deleteBlobsByFilter() -> BlobContainer.listBlobs()... and boom.
Also, AbstractBlobContainer.listBlobsByPrefix() and BlobContainer.deleteBlobsByFilter() can be removed because it has the same drawbacks as AbstractBlobContainer.deleteByPrefix() and also lists all blobs. Listing blobs by prefix can be done at the FsBlobContainer level.
Related to #10344
The static old index tests currently take a long time to run because
each index version essentially recreates the cluster, and spins up
new nodes. This PR instead loads each old version into the existing
cluster as a dangling index. It also removes the intermediate
"StaticIndexBackwardCompatibilityTest" which was an extra layer
with no purpose, and moves a shared version of a commonly found
function to get an http client.
The test now takes between 40 and 60 seconds for me. I also ran it
"under stress" by running all ES tests in one shell, while
simultaneously running 10 iterations of the old index tests. Each
iteration took on average about 90 seconds, which is much better
than the 20+ minutes we see in master on jenkins.
closes#10247
When doc values are explicitly set to the default value serialization
is skipped. This means the alternate way of specifying doc values,
through `fielddata.format: doc_values`, will take precedense if
present.
This change fixes doc values to always be serialized when an explicit value
was passed, so that it continues to take precedence over
`fielddata.format`.
closes#10297closes#10302
The current version is normally a snapshot while in development.
However, when the release process changes the snapshot flag to false,
this causes the static bwc tests to fail because they cannot
find an index for the current version. Instead, this change
skips the current version, because there is no need to test
a verion's bwc against itself.
closes#10292closes#10293
We had an undocumented parameter called `numeric_resolution` which allows to
configure how to deal with dates when provided as a number. The default is to
handle them as milliseconds, but you can also opt-on for eg. seconds.
Close#10072
We recently increased the size of bw indexes and backward compatibility tests
are now taking more time so it makes sense to ask them to do a bit less. This
commit changes the number of replicas we try to copy primaries to from (2 or 3)
to (1 or 2).
Separate repository registration to make sure that failure in registering one repository doesn't cause failures to register other repositories.
Closes#10351
1.1.0 is affected by #5817 which prevents merges from keeping up with the
indexing rate. As a consequence it generates lots of segments and makes bw
compat tests slow. So I added a special case for this version to index fewer
documents.
Many scripts are used to start/stop and install/uninstall elasticsearch. These scripts share a lot of configuration properties like directory paths, max value for a setting, default user etc. Most of the values are identical but some of them are different depending of the platform (Debian-based or Redhat-based OS), depending of the way elasticsearch is started (shell script, systemd, sysv-init...) or the way it is installed (zip, rpm, deb...). Today the values are duplicated in multiple places, making it difficult to maintain the scripts or to update a value.
This pull request make this more uniform: values used in scripts must be defined in a common packaging.properties file. Each value can be overridden in another specific packaging.properties file for Debian or Redhat. All startup and installation scripts are filtered with the common then the custom packaging.properties files before being packaged as a zip/tar.gz/rpm/dpkf archive.
Follow-up of #9135. We initially decreased the stack size because it would end
up a lot of memory when there are many threads. But we also have some thread
pools that may be oversized, in particular the search thread pool.
This commit proposes to decrease the default search thread pool size from
`3 * num_procs` to `3 * num_procs / 2 + 1`. This is large enough to be sure
that we can use all the machine resources even with a search-only work load
but not too large in order to not consume too much memory because of the stack
size and thread locals.
This pull request makes boolean handled like dates and ipv4 addresses: things
are stored as as numerics under the hood and aggregations add some special
formatting logic in order to return true/false in addition to 1/0.
For example, here is an output of a terms aggregation on a boolean field:
```
"aggregations": {
"top_f": {
"doc_count_error_upper_bound": 0,
"buckets": [
{
"key": 0,
"key_as_string": "false",
"doc_count": 2
},
{
"key": 1,
"key_as_string": "true",
"doc_count": 1
}
]
}
}
```
Sorted numeric doc values are used under the hood.
Close#4678Close#7851
A shard recovery response might serialize a shard state at the same time that it is
modified by the recovery process. The test
RelocationTests.testMoveShardsWhileRelocation
failed because of this with a ConcurrentModificationException.
closes#10381
RoutingTables activePrimaryShardsGrouped(), allActiveShardsGrouped() and
allAssignedShardsGrouped() methods treated empty index array input
parameters as meaning "all" indices and expanded to the routing maps
keyset. However, the expansion of index names is now already done in
MetaData#concreteIndices(). Returning an empty index name list here
when a wildcard pattern didn't match any index name could lead to
problems like #9081 because the RoutingTable still expanded this
list of names to "_all". In case of e.g. the recovery endpoint this
could lead to problems.
Closes#9081Closes#10148
This fixes an issue where this was logged:
```
[node_t1] [test][0] flush with org.elasticsearch.action.admin.indices.flush.FlushRequest@65f6f1e
```
by adding a .toString() method to FlushRequest.
It also changes:
```
creating Index [test], shards [1]/[2]
```
to:
```
creating Index [test], shards [1]/[2s]
```
If shadow replicas are being used.
Today we reuse the UUID of the source index on restore. This can create conflicts
with existing shard state on disk. This also causes multiple indices with the same
UUID. This commit preserves the UUID of an existing index or creates a new UUID for
a newly created index.
For quite some time now, our networking layer makes sure to create safe messages as in not using the shared buffers. This is great, and we should remove the old support for "unsafe" notion in our codebase.
closes#10360
Prevents a current edge case resolving concrete aliases or index names in cluster MetaData
that could potentialy lead to NullPointerException when the IndicesOptions don't allow
wildcard expansion and the method is called with aliasesOrIndices argument null or emtpy list.
This change adds a check for that and introduces randomized test that catches this.
Closes#10342Closes#10339
For optimization pruposes a function score query with an empty function
will just result in the original sub query. However, sometimes one might
want to use function_score query to actually filter out docs within for example
bool clauses by using the min_score functionallity.
Therefore the sub query should only be used without wrapping inside
a function_score query if min_score was also not set.
closes#10253closes#10326
When deleting a shard th node that deletes th shard first checks if all shard copies are
started on other nodes. A message is sent to each node tand each node checks locally for
STARTED or RELOCATED.
However, it might happen that the shard is still in state POST_RECOVERY, like this:
shard is relocating from node1 to node2
1. relocated shard on node2 goes in POST_RECOVERY and node2 sends shard started to master
2. master updates routing table and sends new cluster state to node1 and node2
3. node1 processes the cluster state and asks node2 if it has the active shard
before node2 processes the new cluster state (which would cause it to set the shard to started)
4. node2 sends back it does not have the shard started and so node1 does not delete it
This can be avoided by waiting until cluster state that sets the shard to started is actually processed.
closes#10018
Today there is a chance that the state version for shard, index or cluster
state goes backwards or is reset on a full restart etc. depending on
several factors not related to the state. To prevent any collisions
with already existing state files and to maintain write-once properties
this change introductes an incremental state ID instead of using the plain
state version. This also fixes a bug when the previous legacy state had a
greater version than the current state which causes an exception on node
startup or if left-over files are present.
Closes#10316
Now that fine-grained script settings are supported (#10116) we can remove support for the script.disable_dynamic setting.
Same result as `script.disable_dynamic: false` can be obtained as follows:
```
script.inline: on
script.indexed: on
```
An exception is thrown at startup when the old setting is set, so we make sure we tell users they have to change it rather than ignoring the setting.
Closes#10286
The query cache is disabled on dfs_query_then_fetch so we need to enforce
query_then_fetch instead of relying on the randomized search type set by the
test framework.
This commit brings the benefits of the `count` search type to search requests
that have a `size` of 0:
- a single round-trip to shards (no fetch phase)
- ability to use the query cache
Since `count` now provides no benefits over `query_then_fetch`, it has been
deprecated.
Close#7630
Even if there is a background thread that periodically closes search contexts
that seem unused (every minute by default), it is important to close search
contexts as soon as possible in order to not keep unnecessary open files or
to prevent segments from being deleted.
This check would help ensure that refactorings of the SearchContext management
like #9296 are correct.
Adds a getter for the actual netty channel in NettyTransportChannel. The
channel can be used by plugins that need access into netty when processing
requests.
FakeRestRequest is used by a few tests and can also be leveraged by
tests outside of elasticsearch. Moving the package will mean the class
gets exported as part of the test jar.
We already force a refresh in index/create ops, to clear version map
when it's using too much RAM, but we were failing to do this for
deletes, so an app that does tons of deletes with no indexing, and has
set refresh_interval to -1, would have version map using unbounded
RAM.
Closes#10312
After processing mapping updates from the master, we compare the resulting binary representation of them and compare it the one cluster state has. If different, we send a refresh mapping request to master, asking it to reparse the mapping and serialize them again. This mechanism is used to update the mapping after a format change caused by a version upgrade.
The very same process can also be triggered when an old master leaves the cluster, triggering a local cluster state update. If that update contains old mapping format, the local node will again signal the need to refresh, but this time there is no master to accept the request. Instead of failing (which we now do because of #10283, we should just skip the notification and wait for the next elected master to publish a new mapping (triggering another refresh if needed).
Closes#10311
Even if there is a background thread that periodically closes search contexts
that seem unused (every minute by default), it is important to close search
contexts as soon as possible in order to not keep unnecessary open files or
to prevent segments from being deleted.
This check would help ensure that refactorings of the SearchContext management
like #9296 are correct.
When the index service (which holds shards) fails to be created as a result of a shard being allocated on a node, we should fail the relevant shard, otherwise, it will remain stuck.
Same goes when there is a failure to process updated mappings form the master.
Note, both failures typically happen when the node is misconfigured (i.e. missing plugins, ...), since they get created and processed on the master node before being published.
closes#10283
Doc values significantly reduced heap usage, which results in faster
GCs. This change makes the default for doc values dynamic: any
field that is indexed but not analyzed now has doc values. This only
affects fields on indexes created with 2.0+.
closes#8312closes#10209
The local DocumentMapper is updated while parsing and dynamic fields are added before
parsing has finished. If parsing fails after a dynamic field has been added already
then the field was not added to the cluster state but was present in the local mapper of this
node. New documents with the same field would not necessarily cause an update either and
after restarting the node the mapping for these fields were lost. Instead the new fields
should always be updated.
closes#9851closes#9874
We currently have a single bw comp test (FunctionScoreBackwardCompatibilityTests) that requires inline scripts on.
After introducing fine-grained script settings, we moved the internal cluster to use the newer settings, but they are not supported by older nodes started as part of the bw comp tests. Moved script settings out of the default settings, so they won't be part of the ordinary settings when running bw comp tests.
Added logic in FunctionScoreBackwardCompatibilityTests to enable dynamic scripts using the proper setting, depending on the version of the node.
Allow to on/off scripting based on their source (where they get loaded from), the operation that executes them and their language.
The settings cover the following combinations:
- mode: on, off, sandbox
- source: indexed, dynamic, file
- engine: groovy, expressions, mustache, etc
- operation: update, search, aggs, mapping
The following settings are supported for every engine:
script.engine.groovy.indexed.update: sandbox/on/off
script.engine.groovy.indexed.search: sandbox/on/off
script.engine.groovy.indexed.aggs: sandbox/on/off
script.engine.groovy.indexed.mapping: sandbox/on/off
script.engine.groovy.dynamic.update: sandbox/on/off
script.engine.groovy.dynamic.search: sandbox/on/off
script.engine.groovy.dynamic.aggs: sandbox/on/off
script.engine.groovy.dynamic.mapping: sandbox/on/off
script.engine.groovy.file.update: sandbox/on/off
script.engine.groovy.file.search: sandbox/on/off
script.engine.groovy.file.aggs: sandbox/on/off
script.engine.groovy.file.mapping: sandbox/on/off
For ease of use, the following more generic settings are supported too:
script.indexed: sandbox/on/off
script.dynamic: sandbox/on/off
script.file: sandbox/on/off
script.update: sandbox/on/off
script.search: sandbox/on/off
script.aggs: sandbox/on/off
script.mapping: sandbox/on/off
These will be used to calculate the more specific settings, using the stricter setting of each combination. Operation based settings have precedence over conflicting source based ones.
Note that the `mustache` engine is affected by generic settings applied to any language, while native scripts aren't as they are static by definition.
Also, the previous `script.disable_dynamic` setting can now be deprecated.
Closes#6418Closes#10116Closes#10274
In #9893, an enabled flag was added for _field_names. However,
backcompat for indexes created before 1.3.0 (when _field_names
was added) was lost. This change corrects the mapper
to always be disabled when used with older indexes that
cannot have _field_names.
closes#10268
Deleting a type from an index is inherently dangerous because
the type can be recreated with new mappings which may conflict
with existing segments still using the old mappings. This
removes the ability to delete a type (similar to how deleting
fields within a type is not allowed, for the same reason).
closes#8877closes#10231
This adds the Explanation to the explain score again. It is needed
because the explanation of script functions will otherwise not contain
an explanation of _score if boost mode is set to replace.
closes#9826
Fail merge if validate_lat or validate_lon values are not equal. This will prevent inconsistencies between geo_points in a merged index, and parse exceptions for bounding_box and distance filters.
Also merged separate GeoPoint test classes into a single GeoPointFieldMapperTest to be consistent with GeoShapeFieldMapperTests.
closes#10164
* add compiler workarounds for JDK bug JI-9019884
* remove permgen specification during tests (this results in an error on java 9)
* fix threadpool grow/shrink to call methods in the right order (this results in IAE with java 9)
This reverts commit 166fd04239.
Turns out that having log.warn produces a duplicated warn log, as the same message is already logged warn in NettyTranspo
rt#exceptionCaught.
If a request comes in at the same moment the timeout handler for it runs, we may leak a timeoutInfoHolder and erroneously log "Transport response handler not found of id" . The same issue could cause the request tracer to fire a traceUnresolvedResponse call instead of traceReceivedResponse , causing a failure of testTracerLog ( see #10187 ) .
This commit makes sure timeoutInfoHolder is visible before removing the corresponding RequestHolder. It also unifies the TransportService.Adapter#remove(requestId) with TransportService.Adapter#onResponseReceived(requestId), as they are always called together to indicate a response was received.
Closes#10187Closes#10220
When a primary moves to another node, we cancel ongoing recoveries and retry from the primary's new home. At the moment this happens when the primary relocation *starts*. It's a shame as we cancel recoveries that may be close to completion and will finish before the primary has been fully relocated. This commit only triggers the cancelation once the primary relocation is completed.
Next to this, it fixes a race condition between recovery cancellation and the recovery completion. At the moment we may trigger remove a recovered shard just after it was completed. Instead, we should use the recovery cancellation logic to make sure only one code path is followed.
All of the above caused the recoverWhileUnderLoadWithNodeShutdown test to fail (see http://build-us-00.elastic.co/job/es_core_15_debian/32/ ). The test creates an index and then increasingly disallows nodes for it, until only 1 node is left in the allocation filtering rules. Normally, this means we stay in green, but the premature recovery cancellation plus the race condition mentioned above caused a shard to be failed and stay unassigned and the test asserts to fail. This happens due to the following sequence:
- The shard has finished recovering and sent the master a shard started command.
- The recovery is cancelled locally, removing the index shard.
- Master starts shard (deleting it's other copy).
- Local node gets a cluster state with the shard started in it, which cause it to send a shard failed (to make the master aware).
- Shard is failed and can't be re-assigned due to the allocation filter.
The recoverWhileUnderLoadWithNodeShutdown is also adapted a bit to fit the current behavior of allocation filtering (in the past it used to really shut down nodes). Last, all tests in that class are given better names to fit the current terminology.
Clsoes #10218
Today we simply fetch the shards metadata without verifying the
index UUID the shard belongs to. We recently added this UUID
to the shard state metadata. This commit adds verification
to the shard metadata fetching to prevent bringing shards
back into an index it doesn't belong to due to name collisions.
Robert Muir
Ryan Ernst
Julie Tibshirani
Michael McCandless
Martijn van Groningen
Adrien Grand
Simon Willnauer
Lee Hinman
Areek Zillur
Boaz Leskes
Alex Ksikes
Colin Goodheart-Smithe
Christoph Büscher
Jun Ohtani
Shay Banon
Britta Weber
Nicholas Knize
Isabel Drost-Fromm
javanna
David Pilato
Zachary Tong
Leonardo Menezes
jaymode
Tanguy Leroux
Masaru Hasegawa
Igor Motov
Eike Dehling
Simon Willnauer
Ryan Schneider
Reuben Sutton
David Wittman
olivier bourgain
Sebastian Monte
Devin Chollak
Sylwester Lachiewicz