Remove Adjacency_matrix setting #46327
Conversation
|
Pinging @elastic/es-analytics-geo |
|
@jpountz this is the removal of the setting in 8.0 we talked about. Are there additional BWC things I need to add relating to dealing with mixed version clusters or opening old indices? |
| @@ -198,13 +198,13 @@ public String separator() { | |||
| @Override | |||
| protected AggregatorFactory doBuild(SearchContext context, AggregatorFactory parent, Builder subFactoriesBuilder) | |||
| throws IOException { | |||
| int maxFilters = context.indexShard().indexSettings().getMaxAdjacencyMatrixFilters(); | |||
| int maxFilters = SearchModule.INDICES_MAX_CLAUSE_COUNT_SETTING.get(context.indexShard().indexSettings().getSettings()); | |||
There was a problem hiding this comment.
I don't think this is correct since you are looking up a node setting in the index settings?
There was a problem hiding this comment.
UPDATE: @jpountz I got this from QueryParserHelper and it looks like that would have this issue.
However, testing it with an abusive query string/index shows that it does pick up the correct setting from the elasticsearch.yml file so not sure how that magic works its way through.
There was a problem hiding this comment.
Presumably we just substitute for BooleanQuery.getMaxClauseCount() ?
There was a problem hiding this comment.
@jpountz I updated the PR to use BooleanQuery.getMaxClauseCount() as the source of truth. Look good to you?
@romseygeek thought that would be the right thing to do.
markharwood
force-pushed
the
fix/46324
branch
from
b6c32bf
to
911fc0b
Compare
|
(Answering my own question here:) indices created in 7.x with the |
markharwood
force-pushed
the
fix/46324
branch
6 times, most recently
from
784a664
to
f718cb5
Compare
markharwood
force-pushed
the
fix/46324
branch
from
f718cb5
to
e782116
Compare
We dropped the adjacency matrix max clause setting in elastic#46327 because we believe that it is no longer required. From here on out you can use the regular boolean max clause setting.
We dropped the adjacency matrix max clause setting in #46327 because we believe that it is no longer required. From here on out you can use the regular boolean max clause setting.
Now that PR #46257 has improved internal memory usage we no longer need the default limit of 100 filters defined by index.max_adjacency_matrix_filters. This PR removes the setting and uses the existing
indices.query.bool.max_clause_countsetting for Boolean clause limits to apply to the number of filters used in an adjacency matrix.Closes #46324