Restore ElasticSearch functionality

[plenhart]

A couple months ago, had a mediawiki server with 3 wikis running on it dumped in my lap to support. The search box was using CirrusSearch and ElasticSearch / Elastica to generate results, but that was recently broken when the security team did some sort of log4j remediation. All searches would return "An error has occurred while searching: We could not complete your search due to a temporary problem. Please try again later."

So, I removed most traces of CirrusSearch / ElasticSearch and then had the bare-bones MediaWiki search working and of course no longer received the above error. Our server was running back-level MediaWiki (1.30), so I just brought the software up to nearly the latest versions.

Centos 7.3.1611

Product	Version
MediaWiki	1.37.2
PHP	7.4.28 (apache2handler)
MySQL	5.7.20

Extension	Version	Description
CirrusSearch	6.5.4 (70b9a7e) 05:55, 16 March 2022	Elasticsearch-powered search for MediaWiki
Elastica	6.1.3 (c2b259b) 05:46, 19 January 2022	Base Elasticsearch functionality for other extensions by providing Elastica library

# yum list installed|grep elastic
elasticsearch.noarch              6.8.23-1                      @elasticsearch6
elasticsearch-curator.x86_64      5.4.1-1                       installed

I reenabled CirrusSearch and Elastica in LocalSettings.php, and my Version page shows them as loaded, but while I'm still not getting errors, the search results don't seem like they are enhanced in any way. How can I confirm whether or not I'm actually getting results from ElasticSearch instead of the default built-in search? 

I suspect I already have my answer in that my attempt to regenerate my elasticsearch index fails with the response below. So I guess the real question is how do I fix ElasticSearch for my wikis?

# php /data/www/html/wiki/extensions/CirrusSearch/maintenance/UpdateSearchIndexConfig.php
Updating cluster ...
indexing namespaces...
Elastica\Exception\Connection\HttpException from line 190 of /data/www/html/wiki/extensions/Elastica/vendor/ruflin/elastica/lib/Elastica/Transport/Http.php: Couldn't connect to host, Elasticsearch down?
#0 /data/www/html/wiki/extensions/Elastica/vendor/ruflin/elastica/lib/Elastica/Request.php(194): Elastica\Transport\Http->exec()
#1 /data/www/html/wiki/extensions/Elastica/vendor/ruflin/elastica/lib/Elastica/Client.php(689): Elastica\Request->send()
#2 /data/www/html/wiki/extensions/Elastica/vendor/ruflin/elastica/lib/Elastica/Client.php(721): Elastica\Client->request()
#3 /data/www/html/wiki/extensions/Elastica/vendor/ruflin/elastica/lib/Elastica/Index.php(586): Elastica\Client->requestEndpoint()
#4 /data/www/html/wiki/extensions/Elastica/vendor/ruflin/elastica/lib/Elastica/Index.php(330): Elastica\Index->requestEndpoint()
#5 /data/www/html/wiki/extensions/CirrusSearch/includes/MetaStore/MetaStoreIndex.php(501): Elastica\Index->exists()
#6 /data/www/html/wiki/extensions/CirrusSearch/includes/MetaStore/MetaStoreIndex.php(165): CirrusSearch\MetaStore\MetaStoreIndex::cirrusReady()
#7 /data/www/html/wiki/extensions/CirrusSearch/includes/Maintenance/Maintenance.php(221): CirrusSearch\MetaStore\MetaStoreIndex->createIfNecessary()
#8 /data/www/html/wiki/extensions/CirrusSearch/maintenance/IndexNamespaces.php(36): CirrusSearch\Maintenance\Maintenance->maybeCreateMetastore()
#9 /data/www/html/wiki/extensions/CirrusSearch/maintenance/UpdateSearchIndexConfig.php(65): CirrusSearch\Maintenance\IndexNamespaces->execute()
#10 /data/www/html/wiki/maintenance/doMaintenance.php(108): CirrusSearch\Maintenance\UpdateSearchIndexConfig->execute()
#11 /data/www/html/wiki/extensions/CirrusSearch/maintenance/UpdateSearchIndexConfig.php(103): require_once('/data/www/html/...')
#12 {main}

[Skizzerz]

You'll likely need to rebuild all of the elasticsearch indexes, so following the install/upgrade guide from the beginning should be helpful with that. That error indicates that communication can't be established between the wiki and the ES server, so I'd check firewall, etc. for that

[jarmicols]

Yes, UpdateSearchIndexConfig.php failing means either your ES server isn't running/healthy, or CS doesn't know where to find it. If you haven't already you need to set CS configuration parameters in LocalSettings:

$wgCirrusSearchServers = [ 'NAMEOFESSERVER' ];
### Server name can also be an IP or a container name
$wgSearchType = 'CirrusSearch';
### This tells Mediawiki to use CS for searches

A good connection to your ES server will give you script results like:

indexing namespaces...
        Indexing namespaces...done
		content index...
        Fetching Elasticsearch version...6.8.20...ok
        Scanning available plugins...none
        Picking analyzer...english
        Inferring index identifier...minerva_content_first
                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 minerva_content alias...ok
                Validating minerva alias...ok
                Updating tracking indexes...done
...

The best way to confirm if you're getting CS search results is to append '&cirrusDumpQuery' to the search URL for a data-dump in json-format. If you get a json, it's using Cirrus, if you just get the normal search page it isn't.