On 2-May-08, at 11:50 AM, Mike Klaas wrote:
On 2-May-08, at 10:14 AM, Ryan McKinley wrote:
[ ] Keep solr logging as is. (JUL)
[ ] Convert solr logging to SLF4J
[ X ] Abstain
I am not at all part of the java-enterprise-y world, and though as
an outsider it strikes me as odd that that logging implementation in
the language's stdlib is so bad that no-one uses it, that
nevertheless seems to be the case. Though I'm following the
discussion, I'm sticking out of the decision-making process.
For the record, we treat Solr as a stand-alone process, and interact
with it via stderr and signals (log messages are scraped and fed back
into our logging system).
(off-topic-ly, this allows us to easily do things like:
$ tcmd.py -a restart solr
< shutdown 120 Solr servers across the cluster, wait for them to come
back up and register, display any exceptions or timeouts >
)
-Mike
