I increased max connections to 40. I still got: 2011-01-04 21:30:05,701 WARN org.apache.zookeeper.server.NIOServerCnxn: Too many connections from /10.202.50.79 - max is 40 2011-01-04 21:30:06,072 WARN org.apache.zookeeper.server.NIOServerCnxn: Too many connections from /10.202.50.80 - max is 40 2011-01-04 21:30:06,458 WARN org.apache.zookeeper.server.NIOServerCnxn: Too many connections from /10.202.50.80 - max is 40 2011-01-04 21:30:06,944 WARN org.apache.zookeeper.server.NIOServerCnxn: Too many connections from /10.202.50.79 - max is 40 2011-01-04 21:30:07,273 WARN org.apache.zookeeper.server.NIOServerCnxn: Too many connections from /10.202.50.80 - max is 40 2011-01-04 21:30:07,665 WARN org.apache.zookeeper.server.NIOServerCnxn: Too many connections from /10.202.50.79 - max is 40 2011-01-04 21:30:07,876 WARN org.apache.zookeeper.server.NIOServerCnxn: EndOfStreamException: Unable to read additional data from client sessionid 0x12d52be9c2b001b, likely client has closed socket 2011-01-04 21:30:07,876 INFO org.apache.zookeeper.server.NIOServerCnxn: Closed socket connection for client /10.202.50.79:43150 which had sessionid 0x12d52be9c2b001b 2011-01-04 21:30:07,876 WARN org.apache.zookeeper.server.NIOServerCnxn: EndOfStreamException: Unable to read additional data from client sessionid 0x12d52be9c2b008b, likely client has closed socket 2011-01-04 21:30:07,876 INFO org.apache.zookeeper.server.NIOServerCnxn: Closed socket connection for client /10.202.50.79:26104 which had sessionid 0x12d52be9c2b008b 2011-01-04 21:30:07,876 WARN org.apache.zookeeper.server.NIOServerCnxn: EndOfStreamException: Unable to read additional data from client sessionid 0x12d52be9c2b010c, likely client has closed socket
I verified maxClientCnxns of 30 in 0.20.6 where we didn't experience this problem. More comment is welcome. On Tue, Jan 4, 2011 at 11:47 AM, Ted Yu <[email protected]> wrote: > So I should be using HTablePool. > For 0.20.6, I didn't see ConnectionLossException this often. > > I wonder if something changed from 0.20.6 to 0.90 > > On Tue, Jan 4, 2011 at 11:29 AM, Stack <[email protected]> wrote: > >> Are you passing the same Configuration instance when creating your >> HTables? See >> http://people.apache.org/~stack/hbase-0.90.0-candidate-2/docs/apidocs/org/apache/hadoop/hbase/client/HConnectionManager.html<http://people.apache.org/%7Estack/hbase-0.90.0-candidate-2/docs/apidocs/org/apache/hadoop/hbase/client/HConnectionManager.html> >> if not. It explains how we figure whether zk client, rpc connections, >> etc. are shared or not. >> >> St.Ack >> >> On Tue, Jan 4, 2011 at 11:12 AM, Jean-Daniel Cryans <[email protected]> >> wrote: >> > It's a zookeeper setting, you cannot have by default more than 30 >> > connections from the same IP per ZK peer. >> > >> > If HBase is starting ZK for you, do change >> > hbase.zookeeper.property.maxClientCnxns >> > >> > J-D >> > >> > On Tue, Jan 4, 2011 at 11:09 AM, Ted Yu <[email protected]> wrote: >> >> Hi, >> >> I am using HBase 0.90 and our job fails consistently with the following >> >> exception: >> >> >> >> Caused by: org.apache.hadoop.hbase.ZooKeeperConnectionException: >> >> org.apache.zookeeper.KeeperException$ConnectionLossException: >> >> KeeperErrorCode = ConnectionLoss for /hbase >> >> at >> org.apache.hadoop.hbase.zookeeper.ZooKeeperWatcher.<init>(ZooKeeperWatcher.java:147) >> >> at >> org.apache.hadoop.hbase.client.HConnectionManager$HConnectionImplementation.getZooKeeperWatcher(HConnectionManager.java:1035) >> >> ... 19 more >> >> Caused by: >> org.apache.zookeeper.KeeperException$ConnectionLossException: >> >> KeeperErrorCode = ConnectionLoss for /hbase >> >> at >> org.apache.zookeeper.KeeperException.create(KeeperException.java:90) >> >> at >> org.apache.zookeeper.KeeperException.create(KeeperException.java:42) >> >> at org.apache.zookeeper.ZooKeeper.create(ZooKeeper.java:608) >> >> at >> org.apache.hadoop.hbase.zookeeper.ZKUtil.createAndFailSilent(ZKUtil.java:902) >> >> at >> org.apache.hadoop.hbase.zookeeper.ZooKeeperWatcher.<init>(ZooKeeperWatcher.java:133) >> >> ... 20 more >> >> >> >> Zookeeper quorum runs on the same node as NameNode. HMaster is on >> another >> >> node. Hadoop is cdh3b2. >> >> >> >> In zookeeper log, I see (10.202.50.79 is the same node where the >> exception >> >> above happened): >> >> >> >> 2011-01-04 18:47:40,633 WARN org.apache.zookeeper.server.NIOServerCnxn: >> Too >> >> many connections from /10.202.50.79 - max is 30 >> >> 2011-01-04 18:47:41,187 WARN org.apache.zookeeper.server.NIOServerCnxn: >> Too >> >> many connections from /10.202.50.79 - max is 30 >> >> 2011-01-04 18:47:42,375 WARN org.apache.zookeeper.server.NIOServerCnxn: >> Too >> >> many connections from /10.202.50.79 - max is 30 >> >> 2011-01-04 18:47:42,447 WARN org.apache.zookeeper.server.NIOServerCnxn: >> Too >> >> many connections from /10.202.50.79 - max is 30 >> >> 2011-01-04 18:47:43,113 WARN org.apache.zookeeper.server.NIOServerCnxn: >> >> EndOfStreamException: Unable to read additional data from client >> sessionid >> >> 0x12d5220eb970025, likely client has closed socket >> >> 2011-01-04 18:47:43,113 INFO org.apache.zookeeper.server.NIOServerCnxn: >> >> Closed socket connection for client /10.202.50.79:37845 which had >> sessionid >> >> 0x12d5220eb970025 >> >> 2011-01-04 18:47:43,113 WARN org.apache.zookeeper.server.NIOServerCnxn: >> >> EndOfStreamException: Unable to read additional data from client >> sessionid >> >> 0x12d5220eb970087, likely client has closed socket >> >> >> >> Please advise what parameter I should tune. >> >> >> >> Thanks >> >> >> > >> > >
