Atomic mapping updates across types #22220
Conversation
ywelsch
added
:Search Foundations/Mapping
This commit makes mapping updates atomic when multiple types in an index are updated. Mappings for an index are now applied in a single go, which allows to optimize some of the cross-type updates and checks.
ywelsch
force-pushed
the
enhance/atomic-mapping-updates-across-types
branch
from
3d64d56 to
b0bdcbf
Compare
There was a problem hiding this comment.
Thanks, it looks good to me. The one thing I noticed is that some parts of the code that used to raise MapperParsingException now throw an IAE instead. I don't think it is an issue, but I am pinging @bleskes about it since we had a discussion about standardizing those a while back and I'd like to make sure it won't break anything.
| // only update entries if needed | ||
| updatedEntries = internalMerge(indexMetaData, MergeReason.MAPPING_RECOVERY, true, true); | ||
| } catch (Exception e) { | ||
| logger.warn((org.apache.logging.log4j.util.Supplier<?>) () -> new ParameterizedMessage("[{}] failed to apply mappings", index()), e); |
There was a problem hiding this comment.
why are you using the supplier variant here?
There was a problem hiding this comment.
I want to log an exception as well as a parameterized message, so that's the only possible way?
|
The discussion around mapping exceptions we had (if I remember correctly) was around capturing mapping related issues with an incoming document during parsing. This was part of the effort to make failures a valid result of an indexing operation (rather than let exception bubble up to the user). Glancing at the PR, it seems it's all related to mapping merges, so I think we're good on that aspect. |
This commit makes mapping updates atomic when multiple types in an index are updated. Mappings for an index are now applied in a single atomic operation, which also allows to optimize some of the cross-type updates and checks.
This PR makes mapping updates atomic when multiple types in an index are updated. Mappings for an index are now applied in a single atomic operation, which also allows to optimize some of the cross-type updates and checks.