Solr supports two different ways to write v2 APIs: a JSON spec based
approach, and one based on annotated POJOs. The POJO method is now
preferred.
This commit switches the /v2/collections APIs over to the
annotation-based approach. Since V2RequestSupport only works with
jsonspec-based APIs, this commit also changes CollectionAdminRequest
to no longer implement that interface.
This commit introduces a new way for Solr to do backups (with a new
underlying file structure). This new "incremental" backup process
improves over the existing backup mechanism in several ways:
- multiple backups "points" can now be stored at a given backup
location/name, allowing users to choose which point in time they want
to restore
- subsequent backups skip over uploading files that were uploaded by
previous backups, saving time and network time.
- files are checksumed as they're uploaded, ensuring that corrupted
indices aren't persisted and accidentally restored later.
Incremental backups are now the default, and traditional backups
should now be considered 'deprecated' but can still be created by
passing an `incremental=false` parameter on backup requests.
When we are creating a new thread we should give it a descriptive name and enforce this via ForbiddenAPIs. This doesn't apply to Runnable or Callable objects that we pass to an executor, since those should be getting named by the executor itself.
We don't require this in tests because the tests should be more self contained and there is less benefit in descriptive names. If somebody is already profiling a test, then they likely have the context to understand what the unnamed threads are doing, whereas a thread dump from a running Solr instance should have good thread names for everything. This is especially helpful when doing profiling, otherwise we end up with a bunch of Thread-# that are hard to tell apart and search on.
Prior to this commit, SuggestComponent used a HashMap as part of the
response it built on the server side. This class is serialized/
deserialized differently depending on the SolrJ ResponseParser used:
a LinkedHashMap when javabin was used, and a SimpleOrderedMap when XML
was used. This discrepancy led to ClassCastException's in downstream
SolrJ code.
This commit fixes the issue by changing SuggestComponent to avoid these
types that are serialized differently. "suggest" response sections now
deserialize as a NamedList in SolrJ, and the SuggesterResponse POJO has
been updated accordingly.
* When the schema defines _root_, and you want to do atomic/partial updates...
** _root_ needn't be stored or have docValues any more
** _nest_path_ field isn't needed for this any more
** Simplified internal logic
* Allow (and recommend, eventually insist) that the _root_ field be passed for atomic/partial updates to child docs.
** In the absence of _root_, assume the _route_ param is equivalent to ameliorate back-compat scope. This is a temporary hack; remove in SOLR-15064.
** One of the two is required; you'll get an exception if the assumption is false. THIS IS A BACK-COMPAT CHANGE
* Ensure that the update log contains the _root_ field if it's defined in the schema; in some cases it wasn't. It's important for robustness of atomic/partial updates to child docs. Caveat: the buffer replay scenario is not tested with child docs.
* Limited the cases when a realtime searcher is re-opened. It was being applied to any update that included child docs but now only some narrow subset: only for atomic/partial updates, and when the update log contains an in-place update for the same nest because it's complicated to resolve those log entries.
* Internal improvements to RealTimeGetComponent to aid clarity & robustness & probably performance...
** Use SolrDocumentFetcher.solrDoc(docID, ReturnFields) instead of more manual loading. Will do more with this in another PR.
** Clarify when only root doc IDs are expected.
** Use Resolution enum more, add PARTIAL, remove DOC_WITH_CHILDREN; enhance docs.
** When have ReturnFields, a Set of "onlyTheseFields" becomes redundant. Add a child doc resolution via a transformer when needed.
** Clarified where copy-field targets are removed
* NestPathField should default to single valued, instead of inheriting the schema default, which for ancient schemas was multi-valued.
* AddUpdateCommand.getLuceneDocument(s) methods are very internal; made package visible and refactored a bit for clarity
* DocumentBuilder: when in-place update, skip id and _root_ here, thus also simplifying further logic
* NestedShardedAtomicUpdateTest no longer extends AbstractFullDistribZkTestBase because it wasn't really leveraging the "control client" checking, and it added too much complexity to debug failures.
SOLR-13027 Use TestInjection so that we always have a Tragic Event
When we encounter a tragic error in the index writer, we can trigger a
leader election instead of queing up a delete and re-add of the node in
question. This should result in a more graceful transition, and the
previous leader will eventually be put into recovery by a new leader.
closes#2120
Joel Bernstein
Jason Gerlowski
Mike Drob
Andrzej Bialecki
Chris Hostetter
Tim Owen
Noble Paul
Timothy Potter
David Smiley
Tim Dillon
Mike Drob
Ilan Ginzburg
Jan Høydahl
Erick Erickson
Christine Poerschke
Bruno Roustant
Adrien Grand
Dawid Weiss
Michael Aleythe
Tomas Fernandez Lobbe
Noble Paul
Houston Putman
Atri Sharma