[
https://issues.apache.org/jira/browse/SOLR-3424?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=13264730#comment-13264730
]
Uwe Schindler commented on SOLR-3424:
-------------------------------------
The whole design of this cache is wrong:
- it should not be static, the resolved class lookups should only be cached per
instance
- the cache only caches "resolves" not instantiations of encoders, the JVM
caches Class.forName() lookups, so why cache it again?
- To fix the orginal issue, a ConcurrentHashMap would also be fine
- Solr/Lucene use Analyzers which are now enforced to reuse tokenstreams, so
the lookup is only done once when IndexWriter starts and a new indexing thread
is created
The last point tells me: "remove this cache"
> PhoneticFilterFactory threadsafety bug
> --------------------------------------
>
> Key: SOLR-3424
> URL: https://issues.apache.org/jira/browse/SOLR-3424
> Project: Solr
> Issue Type: Bug
> Components: Schema and Analysis
> Affects Versions: 3.6, 4.0
> Reporter: David Smiley
> Assignee: David Smiley
> Priority: Minor
> Fix For: 4.0
>
> Attachments: SOLR-3424_PhoneticFilterFactory_threadsafety_bug.patch
>
>
> PhoneticFilterFactory has a static HashMap registry mapping an encoder name
> to an implementation. There is a ReentrantLock used when the map is modified
> (when the encoder config specifies a class name). However, this map, which
> can be accessed by multiple indexing threads, isn't guarded on any of the
> reads, which isn't just the common path but also the error messages which
> dump the registry into the error message.
> I realize the likelihood of a problem is extremely slim, but a bug's a bug.
--
This message is automatically generated by JIRA.
If you think it was sent incorrectly, please contact your JIRA administrators:
https://issues.apache.org/jira/secure/ContactAdministrators!default.jspa
For more information on JIRA, see: http://www.atlassian.com/software/jira
---------------------------------------------------------------------
To unsubscribe, e-mail: [email protected]
For additional commands, e-mail: [email protected]
