Details
-
Improvement
-
Status: Open
-
Major
-
Resolution: Unresolved
-
None
-
None
-
None
-
None
Description
had a question from "sqwk" on the #solr irc channel last night where he had some questions about weird logs & errors indicating that it wasn't using his solr.xml.
Part of the confusion was SOLR-3264, but i couldn't make sense of the rest.
In talking with miller on IRC today, it occurred to me that file permission problems preventing solr from reading the solr.xml file could explain everything – because unlike trunk, Solr 3.5 didn't log anything special if it couldn't find solr.xml and used the legacy "singlecore" mode as a fallback (an oversight i've already fixed in r1304126)
For many files Solr tries to load, we can't "fail fast" if the file isn't found, or isn't readable, because we support reading from the classpath (and zookeeper) as alternatives, but it would be nice to see if we can come up with a standard way to give good warning/error messages if:
- a file exists, but isn't readable (error?)
- a directory where we are looking for a file exists but isn't readable or executable (warning?)
...i suspect the hardest part of this will be having good test cases