Looks like after 900 seconds, it times out and starts up. I think the
issue is that I'm using the bin/solr start/stop script, and it waits
only 5 seconds before sending a kill -9. In my experience with solr
4.10.x and HDFS, that is not enough time to wait for a large shard to
stop when using HDFS. I've seen it take well over a minute to stop.
I'm not sure if the index is going to be missing data, or if it will be
corrupt at this point.
-Joe
On 4/6/2015 1:35 PM, Joseph Obernberger wrote:
Having a couple issues with restarts of a 27 shard cluster using
SolrCloud 5.0.0 and HDFS. I'm getting errors that a lock file exists
and the shard will not start. When I delete the file, that shard
starts OK.
On another shard, I'm getting the following messsage:
538220 [coreLoadExecutor-5-thread-1] INFO
org.apache.solr.util.FSHDFSUtils รข recoverLease=false, attempt=466 on
file=hdfs://nameservice1:8020/solr5/MAINCOLL/core_node8/data/tlog/tlog.0000000000000002971
after 526067ms
It has been doing this for 526 seconds, and doesn't seem to be coming
up. I've tried restarting it several time, but it seems to be in an
infinite loop retrying. Help!
Thank you.
-Joe
