We currently allow the path of a field alias to be changed through a mapping update. This interacts poorly with percolator queries, as seen in the following gist where changing the path of an alias causes a percolator query to no longer match as expected: https://gist.github.com/jtibshirani/c86b5533b1b4d3ab83fb44280e4817b9. The issue is especially tricky because updating a field alias can update the actual field type (say from keyword to long), which may affect how percolator indexes the query.
This bug raises a larger question: do we want to allow the path of a field alias to change, as is currently implemented?
Relates to #36418, which discusses the possibility of deleting field aliases.
We discussed this issue in the search meeting and agreed to either:
find a way to disallow updates on aliases when percolator fields are in use or disallow using alias field in percolator queries if this can be done in a way that is easy to maintain, or
document this limitation in the percolator docs otherwise.
We currently allow the
path
of a field alias to be changed through a mapping update. This interacts poorly with percolator queries, as seen in the following gist where changing the path of an alias causes a percolator query to no longer match as expected: https://gist.github.com/jtibshirani/c86b5533b1b4d3ab83fb44280e4817b9. The issue is especially tricky because updating a field alias can update the actual field type (say fromkeyword
tolong
), which may affect how percolator indexes the query.This bug raises a larger question: do we want to allow the path of a field alias to change, as is currently implemented?
Relates to #36418, which discusses the possibility of deleting field aliases.