On 8/22/2016 6:20 AM, Chris Rogers wrote: > It’s then that I start seeing lots of errors in the Solr logs, and lots of > repetitive messages appearing in Zookeeper: > > These two Solr errors over and over: > > java.io.IOException: Xid out of order. Got Xid 1299 with err 0 expected Xid > 1298 for a packet with details: clientPath:null serverPath:null > finished:false header:: 1298,14 replyHeader:: 0,0,-4 request:: > org.apache.zookeeper.MultiTransactionRecord@95acc4f3 response:: > org.apache.zookeeper.MultiResponse@0
That appears to be one log message, but you said there were two. Also, this message is incomplete. It is missing the timestamp at the beginning and appears to have been cut off at the end too. I think the message probably had *many* more lines of output that weren't included. > And this from Zookeeper: <snip> > 2016-08-21 21:33:39,147 - WARN > [NIOServerCxn.Factory:0.0.0.0/0.0.0.0:2181:ZooKeeperServer@593] - Dropping > packet at server of type 14 > 2016-08-21 21:33:39,154 - WARN > [NIOServerCxn.Factory:0.0.0.0/0.0.0.0:2181:NIOServerCnxn@634] - > EndOfStreamException: Unable to read additional data from client sessionid > 0x156aefeba2d0004, likely client has closed socket That seems to be saying that Solr closed the connection to zookeeper. I have no idea what might be wrong, based just on what's been provided here. This section of logging seems to contain everything related to the specific connection from port 54548, and if that's true, then it does not appear to have been a timeout. Is there anything in the solr.log file at the timestamp at or near 21:33:39.154(when zookeeper thought the connection was closed)? What version of Solr? What version of zookeeper did you install on the other node? Thanks, Shawn
