After upgrading to 8.x-3.x, our indexing started breaking on particular nodes. After hours of debugging/experimenting, finally got to the bottom of the issue -- a pipe character in a field that was boosted, got added to the boost_term data incorrectly. In that field, the pipe is used to delimit the term from the boost weighting -- but the pipe was added by itself, with no term or weight, and sent to Solr this way.

Solr was throwing an exception, with this in the log:

> Exception writing document id o5zn56-content-entity:node/1412:en to the index; possible analysis error: empty String

... and indexing stopped at this node -- reindexing keeps failing in the same batch so at this point the index becomes stale.

Example in screenshot:

Example of bad pipe character

Comments

freelock created an issue. See original summary.

freelock’s picture

Issue summary: View changes

Fix image in summary.

mkalkbrenner’s picture

confirmed

mkalkbrenner’s picture

Status: Active » Needs review
StatusFileSize
new1.6 KB
mkalkbrenner’s picture

  • mkalkbrenner committed 3d1937f on 8.x-3.x
    Issue #3085299 by mkalkbrenner: Pipe character in a boosted field breaks...
  • mkalkbrenner committed 950b65c on 8.x-3.x
    Issue #3085299 by freelock, mkalkbrenner: Pipe character in a boosted...
mkalkbrenner’s picture

Status: Needs review » Fixed

Status: Fixed » Closed (fixed)

Automatically closed - issue fixed for 2 weeks with no activity.