I'm running 3.5.0 in production (with an old patch from SOLR-1972) and
yesterday's branch_4x in dev (with the most recent SOLR-1972 patch).
Both versions are spitting occasional exceptions. You can see them both
here:
http://pastie.org/private/o2ekh0drs4syqb6t8re4w
I'm pretty sure that the 4.1 exceptions are being triggered when a
recently written homegrown app contacts all my Solr servers (both 3.5
and 4.1) to get /admin/mbeans data from the 16 cores that live on each
server.
The 3.5 server is in production with lots of information in the logs, so
I don't know exactly what's causing it there. I suspect that it's been
doing this for a while now, even before I wrote my /admin/mbeans app.
These exceptions correspond to times when haproxy (using /admin/ping
healthcheck) marks the server DOWN because the health check takes longer
than 5 seconds, and that's been happening for months now, possibly even
for a couple of years.
I already filed SOLR-4191 for the 4.1 side, I just discovered it on 3.5
this morning.
Any ideas?
Thanks,
Shawn
