Yonik Seeley wrote:
On Thu, Jan 29, 2009 at 1:16 PM, Jon Drukman <jdruk...@gmail.com> wrote:
Julian, have you had any luck figuring this out? My production instance
just started having this problem. It seems to crop up after solr's been
running for several hours. Our usage is very light (maybe one query every
few seconds). I saw someone else mention an out of memory error - this
machine has 8GB of RAM and is running 64bit Linux so it's all available to
solr. Our index is very small - under 40MB. the solr process is using
around 615MB of RAM according to top.
I've only seen failure to remove the lock file either when an OOM
exception occured, or the JVM died or was killed.
i guess it's possible that we hit an out of memory error and the
followup lock errors just bumped it out of the log file rotation. i was
running with multilog's default settings so my log files were getting
thrown out very quickly. i just bumped up the JVM's max heap size and
told multilog to keep way more log files so if this happens again
hopefully i will be able to get more info on what happened.
-jsd-
