Jump to content

Extension talk:CirrusSearch/2019

Add topic
From mediawiki.org

Discussion related to the CirrusSearch MediaWiki extension.

See also the open tasks for CirrusSearch on phabricator.

Autocomplete stopped working

[edit]
Latest comment: 7 years ago1 comment1 person in discussion

The following discussion is closed. Please do not modify it. Subsequent comments should be made on the appropriate discussion page. No further edits should be made to this discussion.

EDIT: Js was disabled on the browser!

Autocomplete suddenly stopped working. Paged search results though are provided OK. Checking the indexes with curl localhost:9200/_cat/indices?v gives: 

health status index                                    uuid                   pri rep docs.count     docs.deleted store.size pri.store.size
green  open   wiki_1_31_0_general_1544803724 joVWyY-0QFGNQodiwmnAPw   4   0       1691                9     10.3mb         10.3mb
green  open   wiki_1_31_0_content_1544803501 guv3yKiXQfOMgKTb4f66pA   4   0     583237            97259      5.8gb          5.8gb
green  open   .tasks                                   7EtqCUXhRgqShaTsHeCTMg   1   0          0                0       191b           191b
green  open   mw_cirrus_metastore_first                KWF6MOHuSQOd-d4QCp9zbg   1   0          3                6     11.2kb         11.2kb

Status: 

 elasticsearch.service - Elasticsearch
   Loaded: loaded (/usr/lib/systemd/system/elasticsearch.service; enabled; vendor preset: disabled)
   Active: active (running) since Tue 2019-01-15 22:03:49 EET; 16min ago
     Docs: http://www.elastic.co
 Main PID: 10943 (java)
   CGroup: /system.slice/elasticsearch.service
           └─10943 /bin/java -Xms2g -Xmx2g -XX:+UseConcMarkSweepGC -XX:CMSInitiatingOccupancyFrac...
Spiros71 (talk) 20:16, 15 January 2019 (UTC)Reply
The discussion above is closed. Please do not modify it. No further edits should be made to this discussion.

Install fails while running updateSearchIndexConfig.php - no such index "mw_cirrus_metastore"

[edit]
Latest comment: 7 years ago8 comments6 people in discussion

As the title says, while trying to install CirrusSearch, it fails while running the updateSearchIndexConfig.php script, with the following error: 

Indexing namespaces...[88f74879166e1cb0742b2d96] [no req]   Elastica\Exception\ResponseException from line 179 of /home/mywiki/public_html/w/extensions/Elastica/vendor/ruflin/elastica/lib/Elastica/Transport/Http.php: no such index [index: mw_cirrus_metastore]
Backtrace:
#0 /home/mywiki/public_html/w/extensions/Elastica/vendor/ruflin/elastica/lib/Elastica/Request.php(193): Elastica\Transport\Http->exec(Elastica\Request, array)
#1 /home/mywiki/public_html/w/extensions/Elastica/vendor/ruflin/elastica/lib/Elastica/Client.php(674): Elastica\Request->send()
#2 /home/mywiki/public_html/w/extensions/Elastica/vendor/ruflin/elastica/lib/Elastica/Client.php(706): Elastica\Client->request(string, string, array, array)
#3 /home/mywiki/public_html/w/extensions/Elastica/vendor/ruflin/elastica/lib/Elastica/Index.php(562): Elastica\Client->requestEndpoint(Elasticsearch\Endpoints\DeleteByQuery)
#4 /home/mywiki/public_html/w/extensions/Elastica/vendor/ruflin/elastica/lib/Elastica/Type.php(523): Elastica\Index->requestEndpoint(Elasticsearch\Endpoints\DeleteByQuery)
#5 /home/mywiki/public_html/w/extensions/Elastica/vendor/ruflin/elastica/lib/Elastica/Type.php(491): Elastica\Type->requestEndpoint(Elasticsearch\Endpoints\DeleteByQuery)
#6 /home/mywiki/public_html/w/extensions/CirrusSearch/includes/MetaStore/MetaNamespaceStore.php(65): Elastica\Type->deleteByQuery(Elastica\Query)
#7 /home/mywiki/public_html/w/extensions/CirrusSearch/maintenance/indexNamespaces.php(39): CirrusSearch\MetaStore\MetaNamespaceStore->reindex(LanguageEn)
#8 /home/mywiki/public_html/w/extensions/CirrusSearch/maintenance/updateSearchIndexConfig.php(57): CirrusSearch\Maintenance\IndexNamespaces->execute()
#9 /home/mywiki/public_html/w/maintenance/doMaintenance.php(94): CirrusSearch\Maintenance\UpdateSearchIndexConfig->execute()
#10 /home/mywiki/public_html/w/extensions/CirrusSearch/maintenance/updateSearchIndexConfig.php(73): require_once(string)
#11 {main}

Using Elasticsearch 5.5.2, PHP 7.1.26, MW 1.32.0, Elastica 1.3.0, CirrusSearch 0.2. I've tried multiple versions of Elasticsearch from 5.5.2 up to 5.6.14, all with the same result. --Moydow (talk) 02:24, 19 January 2019 (UTC)Reply

Same error when I run:
<code>php extensions/CirrusSearch/maintenance/updateSearchIndexConfig.php </code>
return this error:
<code>
indexing namespaces...
Indexing namespaces...[432f3d673b42cbf73c70423a] [no req]   Elastica\Exception\ResponseException from line 179 of /var/www/html/extensions/Elastica/vendor/ruflin/elastica/lib/Elastica/Transport/Http.php: no such index [index: mw_cirrus_metastore]
Backtrace:
#0 /var/www/html/extensions/Elastica/vendor/ruflin/elastica/lib/Elastica/Request.php(193): Elastica\Transport\Http->exec(Elastica\Request, array)
#1 /var/www/html/extensions/Elastica/vendor/ruflin/elastica/lib/Elastica/Client.php(674): Elastica\Request->send()
#2 /var/www/html/extensions/Elastica/vendor/ruflin/elastica/lib/Elastica/Client.php(706): Elastica\Client->request(string, string, array, array)
#3 /var/www/html/extensions/Elastica/vendor/ruflin/elastica/lib/Elastica/Index.php(562): Elastica\Client->requestEndpoint(Elasticsearch\Endpoints\DeleteByQuery)
#4 /var/www/html/extensions/Elastica/vendor/ruflin/elastica/lib/Elastica/Type.php(523): Elastica\Index->requestEndpoint(Elasticsearch\Endpoints\DeleteByQuery)
#5 /var/www/html/extensions/Elastica/vendor/ruflin/elastica/lib/Elastica/Type.php(491): Elastica\Type->requestEndpoint(Elasticsearch\Endpoints\DeleteByQuery)
#6 /var/www/html/extensions/CirrusSearch/includes/MetaStore/MetaNamespaceStore.php(65): Elastica\Type->deleteByQuery(Elastica\Query)
#7 /var/www/html/extensions/CirrusSearch/maintenance/indexNamespaces.php(39): CirrusSearch\MetaStore\MetaNamespaceStore->reindex(Language)
#8 /var/www/html/extensions/CirrusSearch/maintenance/updateSearchIndexConfig.php(57): CirrusSearch\Maintenance\IndexNamespaces->execute()
#9 /var/www/html/maintenance/doMaintenance.php(94): CirrusSearch\Maintenance\UpdateSearchIndexConfig->execute()
#10 /var/www/html/extensions/CirrusSearch/maintenance/updateSearchIndexConfig.php(73): require_once(string)
#11 {main}
</code>
Using:
MediaWiki 1.31.1
PHP 7.2.10-0ubuntu0.18.04.1 (apache2handler)
MariaDB 10.1.34-MariaDB-0ubuntu0.18.04.1
ICU 60.2
Elasticsearch 5.6.3
Lua 5.1.5
Elasticsearch is running ok, extensions Elastica 1.3.0.0 (9fcf88c) 08:09 11 oct 2018 and CirrusSearch 0.2 (4330942) 21:17 16 oct 2018
<code>curl -s localhost:9200
{
  "name" : "HHS_8mf",
  "cluster_name" : "docker-cluster",
  "cluster_uuid" : "phiq1MvXRPaSSK5O46vnkQ",
  "version" : {
    "number" : "5.6.3",
    "build_hash" : "1a2f265",
    "build_date" : "2017-10-06T20:33:39.012Z",
    "build_snapshot" : false,
    "lucene_version" : "6.6.1"
  },
  "tagline" : "You Know, for Search"
}
</code>

Hcpag (talk) 09:26, 23 January 2019 (UTC)Reply
same error :( 134.171.183.194 (talk) 14:07, 29 January 2019 (UTC)Reply
same problem here as well.
Environment: Debian Jessie, MW 1.32, JAVA 8, ElasticSearch 5.6.14, CirrusSearch v1.32, Elastica v1.32 209.12.64.2 (talk) 19:53, 31 January 2019 (UTC)Reply
I figured it out. We need to run the metastore.php first, then the updateSearchIndexConfig script. This fixed the problem. The README guide should include this step for new install.
# php metastore.php
mw_cirrus_metastore missing creating.
Creating metastore index... mw_cirrus_metastore_firstScanning available plugins...none
ok
        Green!
Creating mw_cirrus_metastore alias to mw_cirrus_metastore_first.
Update and check the CirrusSearch metastore index. Always operates on a single
cluster.
Usage: php metastore.php [--cluster|--conf|--dbgroupdefault|--dbpass|--dbuser|--dump|--globals|--help|--index-version-basename|--memory-limit|--mwdebug|--profiler|--quiet|--server|--show-all-index-versions|--show-index-version|--update-index-version|--upgrade|--userTestTrigger|--wiki]
Generic maintenance parameters:
    --help (-h): Display this help message
    --quiet (-q): Whether to suppress non-error output
    --conf: Location of LocalSettings.php, if not default
    --wiki: For specifying the wiki ID
    --globals: Output globals at the end of processing for debugging
    --memory-limit: Set a specific memory limit for the script, "max"
        for no limit or "default" to avoid changing it
    --server: The protocol and server name to use in URLs, e.g.
        http://en.wikipedia.org. This is sometimes necessary because server name
        detection may fail in command line scripts.
    --profiler: Profiler output format (usually "text")
    --mwdebug: Enable built-in MediaWiki development settings
Script dependant parameters:
    --dbuser: The DB user to use for this script
    --dbpass: The password to use for this script
    --dbgroupdefault: The default DB group to use.
Script specific parameters:
    --cluster: Perform all actions on the specified elasticsearch
        cluster
    --dump: Dump the metastore index to stdout (elasticsearch bulk index
        format).
    --index-version-basename: What basename to use when running
        --show-index-version or --update-index-version, defaults to wiki id
    --show-all-index-versions: Show all versions for all indices managed
        by this cluster.
    --show-index-version: Show index versions for this wiki.
    --update-index-version: Update the version index for this wiki.
        Dangerous: index versions should be managed by
        updateSearchIndexConfig.php.
    --upgrade: Create or upgrade the metastore index.
    --userTestTrigger: Use config var and profiles set in the user
        testing framework, e.g. --userTestTrigger=trigger
# php updateSearchIndexConfig.php
indexing namespaces...
        Indexing namespaces...done
content index...
        Fetching Elasticsearch version...5.6.14...ok
        Scanning available plugins...none
        Inferring index identifier...mediawiki_content_first
        Picking analyzer...english
        Creating index...ok
                Validating number of shards...ok
                Validating replica range...ok
                Validating shard allocation settings...done
                Validating max shards per node...ok
        Validating analyzers...ok
        Validating mappings...
                Validating mapping...different...corrected
        Validating aliases...
                Validating mediawiki_content alias...alias is free...corrected
                Validating mediawiki alias...alias not already assigned to this index...corrected
                Updating tracking indexes...done
general index...
        Fetching Elasticsearch version...5.6.14...ok
        Scanning available plugins...none
        Inferring index identifier...mediawiki_general_first
        Picking analyzer...english
        Creating index...ok
                Validating number of shards...ok
                Validating replica range...ok
                Validating shard allocation settings...done
                Validating max shards per node...ok
        Validating analyzers...ok
        Validating mappings...
                Validating mapping...different...corrected
        Validating aliases...
                Validating mediawiki_general alias...alias is free...corrected
                Validating mediawiki_archive alias...alias not already assigned to this index...corrected
                Validating mediawiki alias...alias not already assigned to this index...corrected
                Updating tracking indexes...done 209.12.64.2 (talk) 20:58, 1 February 2019 (UTC)Reply
This is supposed to work auto-magically, but apparently is not. I've created https://phabricator.wikimedia.org/T215369 to look into this. EBernhardson (WMF) (talk) 00:56, 6 February 2019 (UTC)Reply
I'm having the same problem here as well.
MediaWiki 1.31.1
PHP 7.1.8 (apache2handler)
MySQL 5.6.10
Elastica 1.3.0.0 (0959e38) 20:02, 24 October 2016
CirrusSearch 0.2 (c23ae6a) 17:21, 25 October 2016
php maintenance/updateSearchIndexConfig.php
indexing namespaces...
Indexing namespaces...done
content index...
Fetching Elasticsearch version...5.6.15...ok
Scanning available plugins...none
Inferring index identifier...mwsandbox_content_first
Picking analyzer...english
Validating number of shards...ok
Validating replica range...ok
Validating shard allocation settings...done
Validating max shards per node...ok
Validating analyzers...ok
Validating mappings...
Validating mapping...ok
Validating aliases...
Validating mwsandbox_content alias...ok
Validating mwsandbox alias...ok
Updating tracking indexes...[c47b9670917f5d1abf7d8a28] [no req] Exception from line 144 of /app/mediawiki/extensions/CirrusSearch/includes/MetaStore/MetaVersionStore.php: meta store does not exist, you must index your data first
Backtrace:
  1. 0 /app/mediawiki/extensions/CirrusSearch/includes/MetaStore/MetaVersionStore.php(69): CirrusSearch\MetaStore\MetaVersionStore->getType()
  2. 1 /app/mediawiki/extensions/CirrusSearch/maintenance/metastore.php(133): CirrusSearch\MetaStore\MetaVersionStore->updateAll(string)
  3. 2 /app/mediawiki/extensions/CirrusSearch/maintenance/metastore.php(95): CirrusSearch\Maintenance\Metastore->updateIndexVersion(string)
  4. 3 /app/mediawiki/extensions/CirrusSearch/maintenance/updateOneSearchIndexConfig.php(290): CirrusSearch\Maintenance\Metastore->execute()
  5. 4 /app/mediawiki/extensions/CirrusSearch/maintenance/updateOneSearchIndexConfig.php(256): CirrusSearch\Maintenance\UpdateOneSearchIndexConfig->updateVersions()
  6. 5 /app/mediawiki/extensions/CirrusSearch/maintenance/updateSearchIndexConfig.php(64): CirrusSearch\Maintenance\UpdateOneSearchIndexConfig->execute()
  7. 6 /app/mediawiki/maintenance/doMaintenance.php(94): CirrusSearch\Maintenance\UpdateSearchIndexConfig->execute()
  8. 7 /app/mediawiki/extensions/CirrusSearch/maintenance/updateSearchIndexConfig.php(73): require_once(string)
  9. 8 {main}
I try to run php maintenance/metastore.php as suggested above.
Update and check the CirrusSearch metastore index. Always operates on a single
cluster.
Usage: php metastore.php [--cluster|--conf|--dbpass|--dbuser|--dump|--globals|--help|--index-version-basename|--memory-limit|--mwdebug|--profiler|--quiet|--server|--show-all-index-versions|--show-index-version|--update-index-version|--upgrade|--userTestTrigger|--wiki]
Generic maintenance parameters:
    --help (-h): Display this help message
    --quiet (-q): Whether to supress non-error output
    --conf: Location of LocalSettings.php, if not default
    --wiki: For specifying the wiki ID
    --globals: Output globals at the end of processing for debugging
    --memory-limit: Set a specific memory limit for the script, "max"
        for no limit or "default" to avoid changing it
    --server: The protocol and server name to use in URLs, e.g.
        http://en.wikipedia.org. This is sometimes necessary because server name
        detection may fail in command line scripts.
    --profiler: Profiler output format (usually "text")
    --mwdebug: Enable built-in MediaWiki development settings
Script dependant parameters:
    --dbuser: The DB user to use for this script
    --dbpass: The password to use for this script
Script specific parameters:
    --cluster: Perform all actions on the specified elasticsearch
        cluster
    --dump: Dump the metastore index to stdout (elasticsearch bulk index
        format).
    --index-version-basename: What basename to use when running
        --show-index-version or --update-index-version, defaults to wiki id
    --show-all-index-versions: Show all versions for all indices managed
        by this cluster.
    --show-index-version: Show index versions for this wiki.
    --update-index-version: Update the version index for this wiki.
        Dangerous: index versions should be managed by
        updateSearchIndexConfig.php.
    --upgrade: Create or upgrade the metastore index.
    --userTestTrigger: Use config var and profiles set in the user
        testing framework, e.g. --userTestTrigger=trigger
elasticsearch
"name" : "IWKNs00",
"cluster_name" : "elasticsearch",
"cluster_uuid" : "ICSITMH-Q9mdgneAfnfjOg",
"version" :
"number" : "5.6.15",
"build_hash" : "fe7575a",
"build_date" : "2019-02-13T16:21:45.880Z",
"build_snapshot" : false,
"lucene_version" : "6.6.1"
"tagline" : "You Know, for Search"
Legaulph (talk) Legaulph (talk) 12:11, 21 February 2019 (UTC)Reply
Upgrade to the version for MediaWiki 1.31 Fixed the issues Legaulph (talk) 17:57, 22 February 2019 (UTC)Reply

Partial matching of page titles

[edit]
Latest comment: 3 years ago5 comments3 people in discussion
Use intitle regex

How can I get results from strings that partially match page titles?

For example: compos* finds instances in titles and text like composition, composed ✓

*mpose finds instances in text, but not in page titles ✘

mpose finds nothing ✘

intitle:mpose finds nothing ✘ Spiros71 (talk) 13:59, 28 January 2019 (UTC)Reply

You can use regex on titles: intitle:/mpose/. DCausse (WMF) (talk) 23:04, 29 January 2019 (UTC)Reply
Sure, but is there some setting so that this is default behaviour? Spiros71 (talk) 07:54, 30 January 2019 (UTC)Reply
I'm wondering the same thing! Would love for this to be default behavior. Frybread (talk) 17:59, 21 April 2023 (UTC)Reply
Unfortunately there are no easy ways to configure CirrusSearch to do this by default, please see my answer at Help talk:CirrusSearch/2022#h-Using_match_phrase_prefix-20220819072000. DCausse (WMF) (talk) 08:29, 24 April 2023 (UTC)Reply

Suggestion: Expose the creation date in search results

[edit]
Latest comment: 7 years ago1 comment1 person in discussion

Issue

The creation date is not exposed in the search results metadata.

Use case

  • As a user, I'd like to assess the age of an article so that I evaluate its potential accuracy.
    • Given that wikis can generally be edited by anyone, the last edit is not indicative of recent improvements because it may simply be vandalism or reversion of said vandalism.
  • As a user, when sorting a page by creation date, I'd like to see the creation date so that it is clear if search results are properly sorted.

Proposed solution

  • Clearly add and label the creation date as well as the last change timestamp; or
  • Make the creation date only visible when the user hovers over the search result.


197.235.235.2 (talk) 09:50, 1 February 2019 (UTC)Reply

Install fails while running updateSearchIndexConfig.php - MetaVersionStore.php: meta store does not exist, you must index your data first

[edit]
Latest comment: 7 years ago4 comments2 people in discussion
RESOLVED
Used the version for 1.31

The following discussion is closed. Please do not modify it. Subsequent comments should be made on the appropriate discussion page. No further edits should be made to this discussion.

MediaWiki 1.31.1 

PHP	7.1.8 (apache2handler)
MySQL	5.6.10
Elastica 1.3.0.0 (0959e38) 20:02, 24 October 2016
CirrusSearch 0.2 (c23ae6a) 17:21, 25 October 2016

I tried

elasticsearch-5.6.15, elasticsearch-5.6.15, elasticsearch-5.6.3

they all result in the same error.

Updating tracking indexes... Exception from line 144 of /MetaVersionStore.php: meta store does not exist, you must index your data first

I started by this

php maintenance/updateSearchIndexConfig.php

check indecis and 

#-> curl -XGET server:9200/_cat/indices?v
health status index                   uuid                   pri rep docs.count docs.deleted store.size pri.store.size
green  open   mwsandbox_content_first JJlWh70ITSG1MJoAI6nCSA   4   0          0            0       162b           162b
yellow open   mw_cirrus_metastore     GSExfBgvRcufIF-TjRIg-A   5   1         35            0      5.9kb          5.9kb

received the above error, Next I cleared everything

run this
php maintenance/metastore.php
mw_cirrus_metastore missing creating.
Creating metastore index... mw_cirrus_metastore_firstScanning available plugins...none
ok
       Green!
Creating mw_cirrus_metastore alias to mw_cirrus_metastore_first.
Update and check the CirrusSearch metastore index. Always operates on a single
cluster.
then
 php maintenance/updateSearchIndexConfig.php
 Updating tracking indexes... Exception from line 144 of /MetaVersionStore.php: meta store does not exist, you must index your data first

curl -XGET server:9200/_cat/indices?v
health status index                     uuid                   pri rep docs.count docs.deleted store.size pri.store.size
green  open   mwsandbox_content_first   ef_Ot4CwQSKguSBO8cqhQg   4   0          0            0       648b           648b
green  open   mw_cirrus_metastore_first WsAVGx39SwuQjt2jFzywrQ   1   0         36            0      9.8kb          9.8kb Legaulph (talk) 13:07, 22 February 2019 (UTC)Reply
I just downloaded CirrusSearch from gitmaster https://github.com/wikimedia/mediawiki-extensions-CirrusSearch/archive/master.tar.gz
Now I get this error
php maintenance/updateSearchIndexConfig.php
indexing namespaces...
Indexing namespaces...done
content index...
Fetching Elasticsearch version...5.6.14...ok
Scanning available plugins...none
Picking analyzer...english
Inferring index identifier...mwsandbox_content_first
[9334359c4b5ee88422d77537] [no req] TypeError from line 35 of /app/mediawiki/extensions/CirrusSearch/includes/Maintenance/IndexCreator.php: Argument 3 passed to CirrusSearch\Maintenance\IndexCreator::__construct() must be of the type array, null given, called in /app/mediawiki/extensions/CirrusSearch/maintenance/updateOneSearchIndexConfig.php on line 336
Backtrace:
#0 /app/mediawiki/extensions/CirrusSearch/maintenance/updateOneSearchIndexConfig.php(336): CirrusSearch\Maintenance\IndexCreator->__construct(Elastica\Index, array, NULL)
#1 /app/mediawiki/extensions/CirrusSearch/maintenance/updateOneSearchIndexConfig.php(320): CirrusSearch\Maintenance\UpdateOneSearchIndexConfig->createIndex(boolean, string)
#2 /app/mediawiki/extensions/CirrusSearch/maintenance/updateOneSearchIndexConfig.php(269): CirrusSearch\Maintenance\UpdateOneSearchIndexConfig->validateIndex()
#3 /app/mediawiki/extensions/CirrusSearch/maintenance/updateSearchIndexConfig.php(64): CirrusSearch\Maintenance\UpdateOneSearchIndexConfig->execute()
#4 /app/mediawiki/maintenance/doMaintenance.php(94): CirrusSearch\Maintenance\UpdateSearchIndexConfig->execute()
#5 /app/mediawiki/extensions/CirrusSearch/maintenance/updateSearchIndexConfig.php(73): require_once(string)
#6 {main}
{
"name" : "IWKNs00",
"cluster_name" : "elasticsearch",
"cluster_uuid" : "ICSITMH-Q9mdgneAfnfjOg",
"version" : {
"number" : "5.6.14",
"build_hash" : "f310fe9",
"build_date" : "2018-12-05T21:20:16.416Z",
"build_snapshot" : false,
"lucene_version" : "6.6.1"
},
"tagline" : "You Know, for Search"
} Legaulph (talk) 13:38, 22 February 2019 (UTC)Reply
I'm afraid that you will run into troubles if you try to run the master branch of CirrusSearch against MW 1.31.
I suggest trying to install CirrusSearch and Elastica using the ExtensionDistributor and the corresponding MW version you use, e.g. https://www.mediawiki.org/w/index.php?title=Special:ExtensionDistributor&extdistname=CirrusSearch&extdistversion=REL1_31
The first error mentions MetaVersionStore.php but this class was not here when 1.31 was cut.
Please let me know if this helps. DCausse (WMF) (talk) 15:40, 22 February 2019 (UTC)Reply
Thank you works perfectly Legaulph (talk) 16:01, 22 February 2019 (UTC)Reply
The discussion above is closed. Please do not modify it. No further edits should be made to this discussion.

Elasticsearch 6.7 supported?

[edit]
Latest comment: 7 years ago6 comments3 people in discussion
RESOLVED
Docu updated by "1.33.x requires ElasticSearch 6.5.x"

The following discussion is closed. Please do not modify it. Subsequent comments should be made on the appropriate discussion page. No further edits should be made to this discussion.

I know the requirements say

> MediaWiki 1.31.x and 1.32.x require ElasticSearch 5.5.x or 5.6.x.

but I'll ask nevertheless ;-)

Will this extension also work with the latest Elasticsearch (6.7), or do I really need to use 5.6? Cboltz (talk) 13:03, 28 March 2019 (UTC)Reply

Good question.
Just last week WMF upgraded Elasticsearch to 6.5.4 so master and 6.5.4 should work. Anyhow, this still leaves the question about 6.7.x. I'd love to see support for this version, too.
In any case the info in the table needs to be extended or amended.
Let's see what the experts say. [[kgh]] (talk) 14:56, 28 March 2019 (UTC)Reply
Unfortunately elasticsearch epitomizes BC hell, it's certainly possible to support both but it's an undertaking. We would accept patches that support both versions, but doing all the legwork for a version that isn't planned to be deployed is unlikely EBernhardson (WMF) (talk) 16:25, 28 March 2019 (UTC)Reply
All fair enough, so to conclude the docu needs to be extended by "MediaWiki 1.33.x requires ElasticSearch 6.5.x."? [[kgh]] (talk) 18:16, 28 March 2019 (UTC)Reply
@EBernhardson (WMF): A quick and dirty confirmation will be nice. Thanks and cheers [[kgh]] (talk) 16:37, 2 April 2019 (UTC)Reply
@EBernhardson (WMF): Obviously we have insulted you badly by asking about the support of ES 6.7.x since you are ignoring us here. I would like to apologize for being rude and giving you a hard time. This was unintentional. [[kgh]] (talk) 17:41, 16 April 2019 (UTC)Reply
The discussion above is closed. Please do not modify it. No further edits should be made to this discussion.

CirrusSearch cannot update live index

[edit]
Latest comment: 7 years ago3 comments2 people in discussion

Hi, All

I have installed CirrusSearch and Elastica, and it works fine with my old wiki pages

But the search result do not update when I add or update a new wiki page.

I have already set $wgDisableSearchUpdate = false after the install

wfLoadExtension( 'Elastica' );
require_once "$IP/extensions/CirrusSearch/CirrusSearch.php";
$wgDisableSearchUpdate = false;
$wgSearchType = 'CirrusSearch';


Anything else I miss?

BTW, my wiki version is below

Product Version
MediaWiki 1.32.1 (7908476)

03:26, 26 March 2019

PHP 7.2.15-0ubuntu0.18.04.2 (apache2handler)
SQLite 3.22.0
Elasticsearch 5.6.16

Jimmysitu (talk) 10:18, 7 April 2019 (UTC)Reply

Do your error logs contain anything related to CirrusSearch? Some job types that are involved in this process are called cirrusSearchLinksUpdate and cirrusSearchElasticaWrite. There might also be logs in a channel called CirrusSearchChangeFailed. EBernhardson (WMF) (talk) 21:44, 8 April 2019 (UTC)Reply
Hi, @EBernhardson (WMF)
Could you help tell me how to check error logs? Jimmysitu (talk) 02:17, 12 April 2019 (UTC)Reply

Searching content of PDF files/PdfHandler-generated text files

[edit]
Latest comment: 7 years ago3 comments1 person in discussion

I think I got CirrusSearch/Elastica/ElasticSearch running and using it as the main search engine. I also have PdfHandler running on MW 1.31.1 and it's able to generate thumbnails and text files of the uploaded PDF. However, I still don't see contents of the PDF in the search results. Is there something I've missed? A configuration setting or something that I need to enable to include the text of the PDF in the search index?

I hope it's okay, I've cross-posted a similar question in Extension talk:PdfHandler. Tommy H (talk) 22:33, 8 May 2019 (UTC)Reply

I'm going through the settings.txt doc right now, but if anyone want to help point me to specifics, it'd be much appreciated. Tommy H (talk) 22:46, 8 May 2019 (UTC)Reply
Okay, not sure what happened, but since I'm running MW on Windows Server (sorry, forgot to mention this before), the standard PdfHandler extension with my "workaround" wasn't working 100%. Thumbnail creation was okay and I thought the pdftotext was working fine, but apparently not. There's a link to a Windows Server version of the PdfHandler extension on the Extension:PdfHandler page.
I used that instead, ran maintenance/update.php, refreshImageMetadata.php, rebuildImages.php as well as extensions/CirrusSearch/maintenance/forceSearchIndex.php as per the https://phabricator.wikimedia.org/source/extension-cirrussearch/browse/master/README file and now it seemed to work and PDF contents are showing up in search results. Tommy H (talk) 02:58, 9 May 2019 (UTC)Reply
[edit]
Latest comment: 5 years ago4 comments4 people in discussion

Hello,

I can’t get CirrusSearch to work. When I run a search, I get this message:

A warning has occurred while searching: $wgCirrusSearchCategoryEndpoint should be set to a valid SPARQL endpoint to use deep category search.


My configuration is :

Product Version
MediaWiki 1.32.0
PHP 7.2.14 (apache2handler)
MySQL 5.7.24
ICU 63.1
Elasticsearch 5.6.16
Extension Version License Description Authors
AdvancedSearch 0.1.0 (9bbb17d) 22:15, 15 October 2018 GPL-2.0-or-later Easy access to advanced search capabilities on Special:Search Thiemo Kreuz, Gabriel Birke, Tonina Zhelyazkova and Christoph Jauera
CirrusSearch 0.2 (b1fa4bd) 13:47, 20 February 2019 GPL-2.0-or-later Elasticsearch-powered search for MediaWiki Nik Everett, Chad Horohoe, Erik Bernhardson and others
Elastica 1.3.0.0 (9fcf88c) 09:09, 11 October 2018 GPL-2.0-or-later Base Elasticsearch functionality for other extensions by providing Elastica library Nik Everett and Chad Horohoe

Thank you for helping me.


Mikael Mikael44115 (talk) 06:55, 19 May 2019 (UTC)Reply

Was anyone able to get this working? I am encountering the same issue. MadX (talk) 23:18, 23 June 2019 (UTC)Reply
I'm getting it too Revansx (talk) 21:51, 8 May 2020 (UTC)Reply
When this error only occurs when trying to specify categories in which the pages should be, as it was the case with me, the following should help:
Disable "deepcat:" functionality in LocalSettings.php as it requires a seperate SPARQL Service.
As specified here, to do that simply add the following line to your LocalSettings.php:
$wgAdvancedSearchDeepcatEnabled = false; // disable deepcat: in favor of incategory: Schlagmichdoch (talk) 04:24, 9 May 2021 (UTC)Reply

Searching on DISPLAYTITLE

[edit]
Latest comment: 7 years ago2 comments2 people in discussion

Hi,

On our private wiki, we removed the restriction to display title due to the types of pages we are producing. Over the last few days I have been trying to modify cirrussearch in order to search on displaytitle. I'm not entirely sure this was possible, but I noticed reference to extracting displaytitle, as well as the schema mentioning it. Has anyone attempted this before? 82.3.50.193 (talk) 09:15, 29 May 2019 (UTC)Reply

This work is tracked in T143396. DCausse (WMF) (talk) 13:26, 29 May 2019 (UTC)Reply

Issue: Can't sort by creation date (create_timestamp_asc )

[edit]
Latest comment: 6 years ago3 comments3 people in discussion

Steps to reproduce:

1. Go to www.mediawiki.org/w/index.php?sort=create_timestamp_asc&search=monkey

Expected: Pages Sorted by ascending Actual Error :"An error has occurred while searching: Query was not understood. Please make it simpler. The query was logged to improve the search system."

Note: This happens on this very wiki, it doesn't seem to affect others. 197.235.55.190 (talk) 12:42, 29 July 2019 (UTC)Reply

It looks like some fields were missing in the search index for mw.org. I've started up a process to fix them, and will need to come up with some process to check the rest of the wikis. It will probably take a few hours before the new index is promoted. EBernhardson (WMF) (talk) 15:21, 29 July 2019 (UTC)Reply
It seems to be working again for this wiki. Thanks. 197.235.66.41 (talk) 21:32, 30 July 2019 (UTC)Reply

How do I enable Completion Suggester?

[edit]
Latest comment: 5 years ago4 comments4 people in discussion

Hi, can anybody help me to enable completion suggester feature in my wiki? The completion suggester options also does not appear in Special:Preferences#mw-prefsection-searchoptions. Any steps that I miss? Thanks for any help. Afiqyazid (talk) 02:15, 2 August 2019 (UTC)Reply

I also struggled to get the suggester working on the latest version. Without knowing your exact setup, this is what worked for me:
LocalSettings.php:
$wgCirrusSearchUseCompletionSuggester = 'yes';
$wgCirrusSearchCompletionSettings = 'fuzzy-subphrases';
$wgCirrusSearchPhraseSuggestProfiles = 'default';
$wgCirrusSearchCompletionSuggesterSubphrases = [
   'build' => true,
   'use' => true,
   'type' => 'anywords',
   'limit' => 10,
];
$wgCirrusSearchCompletionSuggesterUseDefaultSort = true;
You will also need to run <wiki directory>/extensions/CirrusSearch/maintenance/updateSuggesterIndex.php at least once. I ended up setting a cron job to run the updater script nightly. Nirobbins (talk) 16:52, 2 August 2019 (UTC)Reply
For some reason the suggester index was not getting updated through the job queue for language wikis on my wiki family, only for the English wiki were suggestions getting updates. Manually running updateSuggesterIndex.php for the language wiki fixed that, so now I am also creating a cron job to make sure it runs from time to time for each language wiki. Lwangaman (talk) 10:37, 22 September 2020 (UTC)Reply
Using your configuration I got the following error:
Exception caught: Config entry CirrusSearchPhraseSuggestProfiles must be an array or unset",
therefore I just unset $wgCirrusSearchPhraseSuggestProfiles S0ring (talk) 12:42, 5 August 2019 (UTC)Reply

Search backend error during full_text search: number_format_exception

[edit]
Latest comment: 6 years ago3 comments2 people in discussion

Hi,


after the CirrusSearch installation I get the following error (similar like Extension talk:CirrusSearch/2018#h-number_format_exception:_For_input_string:_"0,7"_(solved)-2018-03-23T21:49:00.000Z):

2019-08-02 13:04:55 0292f8cece5d wiki: Search backend error during full_text search for 'science' after 419: number_format_exception: For input string: "0,5"


Here the configuration:

MediaWiki 1.31.3

PHP  7.2.19

MySQL  5.7.26

Elasticsearch  5.6.0

CirrusSearch  0.2

Elastica  1.3.0.0


Is there a fix for this issue?


S0ring (talk) 13:18, 2 August 2019 (UTC)Reply

The problem disappeared after replacing $wgShellLocale from de_DE.utf8 to en_EN.utf8 S0ring (talk) 14:21, 5 August 2019 (UTC)Reply
This issue should have been fixed some time ago, see phab:T189877. DCausse (WMF) (talk) 07:06, 19 August 2019 (UTC)Reply

Running sendData on cluster default 0s after insertion

[edit]
Latest comment: 6 years ago2 comments2 people in discussion

CirrusSearch is installed and running. The following line in ''LocalSettings.php'' records the logging messages:

$wgDebugLogGroups['CirrusSearch'] = "$IP/extensions/CirrusSearch/error.log";


and the following message is frequently recorded:

root@da0488489991:/var/www/html# more extensions/CirrusSearch/error.log

...

2019-08-15 10:46:29 da0488489991 kimmw: Running sendData on cluster default 0s after insertion

2019-08-15 10:46:29 da0488489991 kimmw: Running sendData on cluster default 0s after insertion

2019-08-15 10:46:31 da0488489991 kimmw: Running sendData on cluster default 0s after insertion

2019-08-15 10:46:31 da0488489991 kimmw: Running sendData on cluster default 0s after insertion

2019-08-15 10:46:32 da0488489991 kimmw: Running sendData on cluster default 0s after insertion

...


What does it mean? S0ring (talk) 13:17, 15 August 2019 (UTC)Reply

These are debug messages that you can safely ignore. DCausse (WMF) (talk) 07:10, 19 August 2019 (UTC)Reply

Class 'Elastica\Client' not found

[edit]
Latest comment: 6 years ago4 comments2 people in discussion
RESOLVED
Had to run composer on Elastica directory even though I didn't install from git

The following discussion is closed. Please do not modify it. Subsequent comments should be made on the appropriate discussion page. No further edits should be made to this discussion.

Here's what I'm running:

MediaWiki 1.33.0

PHP 7.2.19-0ubuntu0.18.04.2 (apache2handler)

MySQL 5.7.27-0ubuntu0.18.04.1


I have followed the steps on the extension page:

- Installed Elasticsearch 6.5.4 per the instructions for MediaWiki 1.33.0.

- Installed Elastica, shows up on Special:Version

- Installed CirrusSearch, causes Special:Version to not load after inserting require_once "$IP/extensions/CirrusSearch/CirrusSearch.php"; in LocalSettings.php


The following error is displayed on Special:Version:

[6615ca91ab54427a72fcf3c5] 2019-08-20 14:52:28: Fatal exception of type "Error"


When I attempt to run the first step of the CirrusSearch instructions, this happens:

hincb@cbwiki2:/var/www/html/mediawiki$ php extensions/CirrusSearch/maintenance/updateSearchIndexConfig.php

indexing namespaces...

[c8fe2a117d6f3ed5dab3e395] [no req]   Error from line 90 of /var/www/html/mediawiki/extensions/Elastica/includes/ElasticaConnection.php: Class 'Elastica\Client' not found

Backtrace:

#0 /var/www/html/mediawiki/extensions/Elastica/includes/ElasticaConnection.php(62): ElasticaConnection->getClient()

#1 /var/www/html/mediawiki/extensions/CirrusSearch/includes/Connection.php(133): ElasticaConnection->setConnectTimeout(integer)

#2 /var/www/html/mediawiki/extensions/CirrusSearch/includes/Connection.php(113): CirrusSearch\Connection->__construct(CirrusSearch\SearchConfig, string)

#3 /var/www/html/mediawiki/extensions/CirrusSearch/includes/Maintenance/Maintenance.php(115): CirrusSearch\Connection::getPool(CirrusSearch\SearchConfig, string)

#4 /var/www/html/mediawiki/extensions/CirrusSearch/includes/Maintenance/Maintenance.php(224): CirrusSearch\Maintenance\Maintenance->getConnection()

#5 /var/www/html/mediawiki/extensions/CirrusSearch/maintenance/indexNamespaces.php(35): CirrusSearch\Maintenance\Maintenance->maybeCreateMetastore()

#6 /var/www/html/mediawiki/extensions/CirrusSearch/maintenance/updateSearchIndexConfig.php(54): CirrusSearch\Maintenance\IndexNamespaces->execute()

#7 /var/www/html/mediawiki/maintenance/doMaintenance.php(96): CirrusSearch\Maintenance\UpdateSearchIndexConfig->execute()

#8 /var/www/html/mediawiki/extensions/CirrusSearch/maintenance/updateSearchIndexConfig.php(70): require_once(string)

#9 {main}


Here's verification of my ElasticSearch version:

hincb@cbwiki2:/var/www/html/mediawiki$ curl -X GET "localhost:9200/?pretty"

{

  "name" : "Og7QKNt",

  "cluster_name" : "elasticsearch",

  "cluster_uuid" : "geDUNZn5Tk63F6wjWs8Smg",

  "version" : {

    "number" : "6.5.4",

    "build_flavor" : "default",

    "build_type" : "deb",

    "build_hash" : "d2ef93d",

    "build_date" : "2018-12-17T21:17:40.758843Z",

    "build_snapshot" : false,

    "lucene_version" : "7.5.0",

    "minimum_wire_compatibility_version" : "5.6.0",

    "minimum_index_compatibility_version" : "5.0.0"

  },

  "tagline" : "You Know, for Search"

}


Any ideas why this isn't working? I appear to have the correct versions of everything installed. I've previously configured ElasticSearch 5.4.3 on MediaWiki 1.30.0 and didn't have any issues at all so I generally know what I'm doing until I start getting errors like this. Thanks. Lemonlvor (talk) 15:01, 20 August 2019 (UTC)Reply

Maybe you need to run composer for those dependencies to install correctly, even if installation instructions tell that's only necessary when installing from git Ciencia Al Poder (talk) 16:11, 20 August 2019 (UTC)Reply
It worked perfectly, thanks again! Lemonlvor (talk) 18:18, 20 August 2019 (UTC)Reply
I'll try it. Lemonlvor (talk) 16:26, 20 August 2019 (UTC)Reply
The discussion above is closed. Please do not modify it. No further edits should be made to this discussion.

CirrusSearch not working, nor meeting bugs

[edit]
Latest comment: 6 years ago3 comments3 people in discussion
On my machine, ElasticSearch is installed.
In Special:Version, it shows that both Elastica and CirrusSearch have been installed already.
However, clicking the "(?)Help" sign beside the search bar leads me to "Help:Searching" instead of "Help:CirrusSearch," and everything shows that I can't use any of functions of CirrusSearch, except for the "Special:Version" page.
What might be the hidden problem? Haven't I correctly installed preparations and the CirrusSearch module? Neil9830409 (talk) 03:10, 2 September 2019 (UTC)Reply
You might want to check your CirrusSearch extension version. I ran into the same problem after upgrading from Ubuntu 18.04 LTS to 19.04 and Mediawiki 1.27 to 1.31 on my quest to set up PDF indexing and search support. The CirrusSearch installation recommendation for non-supported versions of Mediawiki (1.27 on Ubuntu 18.04 LTS) was to try the extension for the latest Mediawiki version available (1.33). Ubuntu 19.04 ships with Mediawiki 1.31. After installing the correct CirrusSearch extension version (1.33 -> 1.31) it started working like a charm. TeeDizzle (talk) 04:07, 4 September 2019 (UTC)Reply
On my wiki, the help link still points to Help:Searching, however Cirrus Search is fully functional with all its features.
Does a simple search work? Please elaborate on the results you get Ciencia Al Poder (talk) 09:15, 2 September 2019 (UTC)Reply

Newer elasticsearch with mw 1.34?

[edit]
Latest comment: 6 years ago2 comments2 people in discussion

Will MediaWiki 1.34 use an updated version of ElasticSearch? Currently, 6.5.x is supported, which came out in December 2019. It would be great to support 7.x series, but even a bump to 6.8.x series would be amazing.

Is there a roadmap somewhere? TazzyTazzy (talk) 19:49, 26 September 2019 (UTC)Reply

There is no particular roadmap for elasticsearch versions. Unfortunately elasticsearch is an ever moving target, and upgrades take a significant (multiple month) time investment. Currently I'm not aware of anything in the later elasticsearch versions that make that investment worthwhile. EBernhardson (WMF) (talk) 21:30, 4 October 2019 (UTC)Reply

Suggestion: Make it possible to only search within certain slots

[edit]
Latest comment: 6 years ago1 comment1 person in discussion

Use cases:

Searching only content of pdf files (assuming it ever gets moved to a separate slot).

Exclude text from structured data:

Currently searching for "P180" in commons finds (c:Special:Search/P180 ) stuff from a different slot that isn't even included in the page source nor any of its templates.

Suggested solution:

A keyword to exclude or include slots, e,g, "slot:wikitext", "slot:filedata", "slot:structured/ mediainfo".


Currently, insource serves as a reasonable workaround, but will fail in cases where the content come from a template, for instance in the P180 case above, one could try doing something like ' "P180" -insource:P180' or ; 'Academia argentina muestran los primeros 5 -insource:"Academia argentina muestran los primeros 5" ' . But that would exclude pages that include the search string in both wikitext and file contents. 197.235.220.190 (talk) 20:19, 17 October 2019 (UTC)Reply

CirrusSearch search ranking

[edit]
Latest comment: 6 years ago1 comment1 person in discussion

My wiki is version 1.34-rc.0 in language zh-hk with CirrusSearch installed.

Recently I have upgraded the elasticsearch from 5.X to 6.8.4, and the following annoying happens.

Selected "sort by relevance", when I search 5月1日 in the search bar, the first one in the list is the article with title 5月1日, but the following are just the combinations of the 4 words, such as 1月5日, 5月 and 5月5日 (See this); but if I type the keyword with the double-quote, i.e. "5月1日", it shows all the articles which contains this phrase 5月1日 in the text. (See this)

My project team prefers the second result without typing the double-quote, how should I set the config? Thanks a lot!

MediaWiki 1.34.0-rc.0
Elasticsearch 6.8.4
Elastica 6.0.2 (bf50f87)
CirrusSearch 6.5.4 (68073ed)

Eagle (talk) 18:37, 30 October 2019 (UTC)Reply

forceSearchIndex.php failing on some batches

[edit]
Latest comment: 6 years ago2 comments1 person in discussion

When I run the forceSearchIndex.php --skipLinks --indexOnSkip maintenance script to initiate the indexes (after clearing them with updateSearchIndexConfig.php --startOver) I'm finding that some batches (not always the same ones) are failing with the exception below. It seems to be load related because if I run the script over 5K articles at a time there's no problem, but I can't run thru them all in one go even adjusting --batchSize or --pauseForJobs etc. Also I noticed that the second pass using --skipParse can run thru all articles without problem.

We have MW versions 1.31 - 1.33 and its happening on them all. We're using the redis job queue.

Anyone have any idea what might be happening here?

2019-11-18T11:39:54-0400 ERROR: Runner loop 1 process in slot 3 gave status '255':
json_decode() error (4): Syntax error
php /var/www/dcs/htdocs/wiki/maintenance/runJobs.php --wiki='dcs-nrgdebt_' --type='cirrusSearchMassIndex' --maxtime='60' --memory-limit='300M' --result=json STDERR:
	PHP Fatal error:  Uncaught Wikimedia\Rdbms\DBTransactionError: Wikimedia\Rdbms\LBFactory::shutdown: transaction round 'CirrusSearch\Job\MassIndex::run' still running. in /var/www/dcs/htdocs/wiki/includes/libs/rdbms/lbfactory/LBFactory.php:258
Stack trace:
#0 /var/www/dcs/htdocs/wiki/includes/libs/rdbms/lbfactory/LBFactory.php(183): Wikimedia\Rdbms\LBFactory->commitMasterChanges('Wikimedia\\Rdbms...')
#1 /var/www/dcs/htdocs/wiki/includes/libs/rdbms/lbfactory/LBFactory.php(156): Wikimedia\Rdbms\LBFactory->shutdown(0)
#2 /var/www/dcs/htdocs/wiki/includes/libs/rdbms/lbfactory/LBFactory.php(734): Wikimedia\Rdbms\LBFactory->destroy()
#3 [internal function]: Wikimedia\Rdbms\LBFactory->__destruct()
#4 {main}
  thrown in /var/www/dcs/htdocs/wiki/includes/libs/rdbms/lbfactory/LBFactory.php on line 258

Nad (talk) 00:25, 20 November 2019 (UTC)Reply

p.s. the json_decode error is just from the job runner script because the json contains the exception text, so the problem has already occurred at that point. Nad (talk) 00:28, 20 November 2019 (UTC)Reply

Suggestion: Provide descriptive labels for certain keywords

[edit]
Latest comment: 6 years ago1 comment1 person in discussion

Issue:

Keywords such as "inlanguage:" do not provide appropriate feedback when the keyword input is entered, and the input itself is cryptic.

For example, one may expected that "inlanguage:jp" is an appropriate term for the japanese language, considering that wikimedia sites themselves redirect it to japanese wikipedia ( jp:).


Proposed solution:

Add a simple description that expands or makes it clearer that the search has succeeded and provides expanded terms whenever appropriate, e.g.:

Showing search results for "Pages in the japanese (ja) language "


This also affects the Extension:WikibaseCirrusSearch , see Extension talk:WikibaseCirrusSearch#h-Suggestion:_Provide_labels_for_wikibasecirrussearch_(e.g._haswbstatement)-2019-12-09T21:58:00.000Z . 197.218.84.165 (talk) 22:10, 9 December 2019 (UTC)Reply

Error using deepcat

[edit]
Latest comment: 5 years ago4 comments4 people in discussion

I posted this to the AdvancedSearch extension as well, but I thought I'd try here, too:

Product Version
MediaWiki 1.33.0
PHP 7.2.24-0ubuntu0.18.04.1 (fpm-fcgi)
MySQL 5.7.28-0ubuntu0.18.04.4
ICU 60.2
Elasticsearch 6.5.4
AdvancedSearch 0.1.0 (76eea63) 16:40, 25 March 2019 GPL-2.0-or-later Easy access to advanced search capabilities on Special:Search Thiemo Kreuz, Gabriel Birke, Tonina Zhelyazkova, Christoph Jauera, Kai Nissen and Tim Eulitz
CirrusSearch 0.2 (2daa9b8) 09:07, 25 March 2019 GPL-2.0-or-later Elasticsearch-powered search for MediaWiki Nik Everett, Chad Horohoe, Erik Bernhardson and others
Elastica 6.0.2 (eee38b0) 07:37, 10 September 2019 GPL-2.0-or-later Base Elasticsearch functionality for other extensions by providing Elastica library Nik Everett and Chad Horohoe

I'm sure this is something I just haven't configured correctly, but can someone point me to how this should be configured?

I'm trying to search multiple categories with Advanced Search. Using the default setup in CirrusSearch, which appears to be

$wgCirrusSearchCategoryEndpoint = '';

and getting the following message:

$wgCirrusSearchCategoryEndpoint should be set to a valid SPARQL endpoint to use deep category search.

So it makes sense that I'm getting the error, I just don't know what this endpoint should look like.

Any assistance? Blinkingline (talk) 18:46, 11 December 2019 (UTC)Reply

Looks like this requires Wikibase according to phab:T192942.
Why a local-wiki category search requires an entire wikibase instance? I don't know, but that doesn't look reasonable. Ciencia Al Poder (talk) 10:20, 12 December 2019 (UTC)Reply
Deepcat requires both a blazegraph instance and regular dumps exported from your wiki to that blazegraph instance. Overall I do not think running deepcat outside WMF infrastructure is going to be satisfying, and will instead be a giant headache of trying to run additional complex services that generally only make sense if you already have those services for other purposes.
In theory the same thing could be implemented with recursive SQL queries without using blazegraph, but we would not deploy recursive SQL queries due to performance considerations. If a patch were provided to implement this slow-path we can review it for inclusion and use outside WMF. EBernhardson (WMF) (talk) 16:43, 12 December 2019 (UTC)Reply
When this error only occurs when trying to specify categories in which the pages should be, as it was the case with me, the following should help:
Disable "deepcat:" functionality in LocalSettings.php as it requires a seperate SPARQL Service.
As specified here, to do that simply add the following line to your LocalSettings.php:
$wgAdvancedSearchDeepcatEnabled = false; // disable deepcat: in favor of incategory: Schlagmichdoch (talk) 04:26, 9 May 2021 (UTC)Reply

error download

[edit]
Latest comment: 6 years ago2 comments2 people in discussion

extdist.wmflabs.org/dist/extensions/CirrusSearch-REL1_34-a86e0a5.tar.gz


404

nginx/1.13.9 Elserolf (talk) 16:45, 29 December 2019 (UTC)Reply

I've reported the issue on phab:T241542 Ciencia Al Poder (talk) 18:29, 29 December 2019 (UTC)Reply
Retrieved from "https://www.mediawiki.org/w/index.php?title=Extension_talk:CirrusSearch/2019&oldid=7505262"