we’ve recently discovered a memory leak in the
PerFieldAnalyzer. If you use the PerFieldAnalyzer
(which you acutally should), you should switch
to a pure ruby version of that analyzer. The C
version of the Analyzer is consuming memory
on every analyzing request.
You can find an example script to verify the
leak[1]. Furthermore we’ve added a
workaround, building a PerFieldAnalyzer
in pure ruby[2]. You can read more about the
leak on the omdb tech blog[3].
we’ve recently discovered a memory leak in the
PerFieldAnalyzer. If you use the PerFieldAnalyzer
(which you acutally should), you should switch
to a pure ruby version of that analyzer. The C
version of the Analyzer is consuming memory
on every analyzing request.
Is this relevant to those of us who do not explicitly use
PerFieldAnalyzer? i.e. is it used behind the scenes unless you say
otherwise?