Commit Graph

2040 Commits

Author SHA1 Message Date
kimchy fcdd2ba11c And/Or Filter: Possible failure when inner filter does not match any docs, closes #892. 2011-04-29 21:45:00 +03:00
kimchy 887ac1fce6 Search: Optimize (perf) execution of global facets, closes #889. 2011-04-28 19:32:13 +03:00
kimchy 2c0bb9199a Search: Optimize (perf) execution of global facets, closes #889. 2011-04-28 17:20:31 +03:00
kimchy 0ab8d1f414 Search API: filter element should not be applied to global facets, closes #888. 2011-04-28 16:16:50 +03:00
kimchy 19448d7775 include_in_all not overridable in objects, closes #883. 2011-04-28 15:44:25 +03:00
kimchy 5c8fe598a6 Scripting - mvel: Fix thread safety misuse when compiling scripts, first go it so use ParserConfiguration, second is a fix in mvel itself, relates to #886. 2011-04-27 00:38:51 +03:00
kimchy 6ecbef463f Create Index API: Better logic to wait for ack for indices being created on nodes, closes #879. 2011-04-27 00:24:27 +03:00
kimchy 6ace8ba0d8 Query DSL: Allow to set boost on has_child query, closes #884. 2011-04-26 20:40:21 +03:00
kimchy f3e032315f (Small) optimization to has_child filter, closes #880. 2011-04-25 17:37:56 +03:00
kimchy 49ac30801e more cache cleaning 2011-04-25 00:16:42 +03:00
kimchy 23c17d7101 improved close logic on index query parser 2011-04-24 23:42:59 +03:00
kimchy 013c4e9fa4 improved cleaning of cached parser context 2011-04-24 23:34:59 +03:00
kimchy 990f371580 better cached streams logic 2011-04-24 23:28:35 +03:00
kimchy 9b262a7363 clean teh lzf buffer recycler 2011-04-24 22:48:38 +03:00
kimchy 1e5dc09848 Unexpected failure to create a shard can lead to data loss if it has no replicas, closes #878. 2011-04-23 13:29:14 +03:00
kimchy 3b07f77f3d clean indices when created just for mapping processing 2011-04-22 16:34:54 +03:00
kimchy 65e05538f0 refresh changed mapping in cluster metadata, this will happen when upgrading from 0.15 to 0.16 2011-04-22 16:07:18 +03:00
kimchy 0319972d77 change logging level to debug 2011-04-22 04:43:23 +03:00
kimchy 6ed2624772 fix wrong closing of possible null mappings in geo point mapping 2011-04-22 04:29:47 +03:00
kimchy d517c3dfef upgrade to jackson 1.8.0 2011-04-22 03:45:56 +03:00
kimchy e171596a6f default replacement char to delimiter char 2011-04-22 03:20:34 +03:00
kimchy 1f89fb26aa upgrade to mvel 2.1.beta2 (not to be confused with Beta2 :) ) 2011-04-22 02:33:39 +03:00
kimchy d8d47abcba no need to cache filters again when handling named filters 2011-04-22 00:37:20 +03:00
Igor Motov 4c295a28d5 Don't attempt to load camelCased version of the package name.
It's unlikely that a package would have a camelCased name and an attempt to load a class with wrong case is causing NoClassDefFoundError instead of ClassNotFoundException on non case-sensitive file systems.
2011-04-21 21:39:17 +03:00
kimchy db04649271 Search request intermittent failures with has_child query/filter, closes #875. 2011-04-21 17:54:01 +03:00
Lukáš Vlček adc3dc0e99 Fix for highlighting when number_of_fragments is set to 0 and term_vector is not set, closes #873 2011-04-21 14:55:21 +02:00
kimchy dedbedbb77 Update Settings: Allow to dynamically set filter cache settings, closes #874. 2011-04-21 14:40:57 +03:00
Lukáš Vlček add6db352d Added test for the highlighting using the source field. (This test probably dropped out in one of previous commits) 2011-04-21 11:55:35 +02:00
kimchy 568dec6b5a Simplify Filter Cache Eviction Settings, closes #871. 2011-04-21 02:05:05 +03:00
kimchy 90b88fe4b7 add TODO optimizations comments 2011-04-21 01:57:28 +03:00
kimchy 9c606c7b1d better caching of LZF streamns 2011-04-20 13:37:30 +03:00
kimchy f5dbcb2fa4 Failure to recover a shard might cause loosing translog data (especially with no replicas), closes #869. 2011-04-20 04:52:47 +03:00
kimchy 3c233347b8 make sure we keep order in mappings when recovering so there won't be reparsing of the mappings on nodes (introduced with faster recovery work done on gateway) 2011-04-20 03:42:51 +03:00
kimchy af39f07213 Mapping: By default, don't index _id, closes #868. 2011-04-20 02:49:14 +03:00
kimchy 3b21759bec Better support with _type is marked as not indexed, allowing to filter by type, closes #866. 2011-04-20 01:31:38 +03:00
kimchy c3f3c268c8 Query DSL: Add ids query and filter to fetch docs by ids (do not require _id field to be indexed), closes #865. 2011-04-20 00:30:30 +03:00
Lukáš Vlček be2a456cc0 improve test: check hits length and zero failures 2011-04-19 21:49:08 +02:00
kimchy 8e027b3baf properly handle cases where filter returns null docIdSet with and/or/not filters 2011-04-19 21:28:08 +03:00
kimchy 5cc943c7db remove some field caching logic, does not seem to do much perf wise, and simplifies back the code 2011-04-19 17:07:32 +03:00
kimchy 56c9783cb3 remove using field cache when indexing string fields for now 2011-04-19 16:16:31 +03:00
kimchy f8ff42d397 log when failing to process doc after index operation 2011-04-19 16:13:26 +03:00
kimchy 79ebcc31c5 double check that there are mappings before trying to call the post process method 2011-04-19 16:10:51 +03:00
kimchy 32127c4d73 register the future with teh listener when deleting an index, so it can be canceled 2011-04-19 15:25:02 +03:00
kimchy e40499f35f small improvement to bulk indexing to not allocated ops if no percolation is done 2011-04-19 13:53:21 +03:00
kimchy 3b82b013d5 eagerly reroute shards when opening an index 2011-04-19 12:56:08 +03:00
kimchy ffc8c3b68a Weird error message on syntax error, closes #864. 2011-04-19 12:23:29 +03:00
kimchy d29754121c don't log a versioning conflict failure, even in DEBUG level 2011-04-19 12:17:11 +03:00
kimchy b94fab34eb Mapping: Allow to set index to `no` for _id and _type, closes #863. 2011-04-17 21:06:46 +03:00
kimchy c4be4975d6 better message 2011-04-17 19:00:45 +03:00
kimchy c2ee6dd120 improve gateway recovery and applying indices, also improve speed of index creation 2011-04-17 18:39:00 +03:00
kimchy 287dc862e0 more options on stress test 2011-04-17 10:14:49 +03:00
kimchy 166493f9d5 properly clean mapper data, and cache some mapping structures 2011-04-17 09:52:29 +03:00
kimchy efb3e97ce4 don't return any id when doesn't exists 2011-04-17 06:21:39 +03:00
kimchy fb35b1c993 clean guice caches after creating injectors so it will take less memory 2011-04-17 05:16:34 +03:00
kimchy 343c80b100 shard tokenizers, token filters, char filters across indices / shards 2011-04-17 03:55:33 +03:00
kimchy 067b8379d5 internal cluster name 2011-04-17 01:45:02 +03:00
kimchy 4907ce4fec intern logger names 2011-04-17 01:20:24 +03:00
kimchy c51b481cc9 revert back to guice 2 2011-04-17 00:58:00 +03:00
kimchy 73e7ea4663 allow to control index indexing concurrency (pretty much internal) 2011-04-16 22:55:12 +03:00
kimchy e8306ac2c8 improve local gateway allocation by caching the started shards when possible and not asking the nodes for it 2011-04-16 22:33:45 +03:00
kimchy 658594fa70 improve gateway initial recovery time by creating the indices, and only after all have been created, do the rerouting 2011-04-16 18:23:48 +03:00
kimchy 763f986a30 Versioning: Better handling of deletes - time based eviction, closes #862. 2011-04-16 17:26:31 +03:00
kimchy 4eddaec8ba change defaults in stress test 2011-04-16 16:57:17 +03:00
kimchy a1796c3408 add an estimated time thread and use it where we don't need exact time 2011-04-16 16:57:02 +03:00
kimchy 2cc4a286f8 small refactor 2011-04-16 03:23:47 +03:00
kimchy 8fa8a72175 add refresh before verifying in the test 2011-04-16 03:20:52 +03:00
kimchy 9927fd4cdd fix none bloom cache to proper name 2011-04-16 03:15:43 +03:00
kimchy 5ab0ec4911 add a concurrent indexing stress test 2011-04-16 03:15:13 +03:00
kimchy 9c85500246 change additionalSettings to use Settings 2011-04-16 01:20:27 +03:00
Igor Motov 785fb11f2a Allow plugins to provide additional settings. 2011-04-16 01:17:23 +03:00
kimchy bba7179eeb automatically clean indices that no longer exists in the metadata, this can happen if we delete a closed index, or when a node joins a cluster, that no longer has those indices 2011-04-14 21:15:55 +03:00
kimchy 897587f981 Allow to delete a closed index, closes #859. 2011-04-14 20:51:20 +03:00
kimchy 7c38f20056 Query DSL: Geo Distance Range filter, closes #856. 2011-04-14 20:20:18 +03:00
kimchy 7874291c0e fix wrong initialization of mem evictions counter 2011-04-14 19:13:14 +03:00
kimchy 1e84c439e5 Shard Allocation: Closed indices are not properly taken into account when rebalancing, closes #858. 2011-04-14 15:00:54 +03:00
kimchy 7550fc570e Query DSL: Allow to directly wrap a query with a constant_score query, closes #857. 2011-04-14 14:19:24 +03:00
kimchy e5c1ab0ca6 add another reroute method to do no assignment but still elect new primaries on failed nodes 2011-04-14 13:43:01 +03:00
kimchy 25eba4b60d Unicast Discovery: Concurrently connect to nodes to improve cases where some listed nodes are not up, closes #855. 2011-04-14 02:27:59 +03:00
kimchy 3b72d63035 EC2 Zen Discovery: Automatically use the configured transport port to ping other nodes, closes #854. 2011-04-14 02:13:27 +03:00
kimchy 853ade7297 Upgrade to jackson 1.7.6 (fixes that long field name smile bug) 2011-04-14 01:36:19 +03:00
kimchy 65fde3e71e only remove shards if they have been started on all other nodes 2011-04-14 01:12:37 +03:00
kimchy 22b5e7ded8 Local Gateway: Listed active shards on each node does apply index deletion, closes #853. 2011-04-13 16:15:26 +03:00
kimchy 41af86dd7e internal option ot pretty print state 2011-04-13 16:04:18 +03:00
kimchy b80324db08 only serialize the cluster state ones when publishing it to multiple nodes 2011-04-13 15:36:28 +03:00
kimchy e9bc7f7d95 River: Failure to reallocate river to another node on rivers node failure, closes #850. 2011-04-12 21:52:02 +03:00
kimchy e8503c1455 Better handling when a shard hits OOM, closes #848. 2011-04-11 21:10:56 +03:00
kimchy decb5fa898 Optimize API: Don't execute concurrent optimize operations (shard level) on the same node, closes #846. 2011-04-10 23:12:00 +03:00
kimchy 7d8726a5e8 Better handling of shard failures, closes #845. 2011-04-10 21:59:59 +03:00
kimchy 8eab5ec528 experiment with lab for filter cache, disabled by default for now, also, move to use soft reference on the whole reader key value, and not per filter (less load) 2011-04-09 20:21:23 +03:00
kimchy b658fba22e allow to copy openbitset into a sliced one 2011-04-09 14:33:16 +03:00
kimchy 991683efa6 remove clearUnreferenced from caches, no longer need it 2011-04-09 14:20:18 +03:00
kimchy 49e2f55abd actually register the eviction listener with soft / resident filter cache 2011-04-09 14:11:42 +03:00
kimchy c934f04b0c remove weak caching of filters and stick with only caching when needed / flagged / defaulted 2011-04-09 02:47:31 +03:00
kimchy b71513a072 JMX: Don't use node id and name as part of the jmx registration, use static org.elasticsearch, closes #843. 2011-04-08 18:49:40 +03:00
kimchy 388f4febba add more terms stats bench 2011-04-07 20:41:20 +03:00
kimchy 93372868a2 fix error when trying to delete all indices and none exists 2011-04-07 20:08:53 +03:00
kimchy ca01dc7a09 Network Settings: Allow to explicitly set ipv4 and ipv4 when using _networkInterface_ notation, closes #841. 2011-04-07 20:01:52 +03:00
kimchy 1fdef91fc2 Allow to search across indices and types even if some types do not exists on some indices, closes #839. 2011-04-07 16:00:18 +03:00
kimchy 4603d10501 faster handling of a case where each doc has multi valued values for different facets, also, don't take into account missing value values (and in the future, we can also count those per bucket if needed) 2011-04-07 15:04:54 +03:00
kimchy b479bc58f9 faster handling of a case where each doc has multi valued values fields in histogram facet 2011-04-07 14:52:25 +03:00