Thanks. Is it possible to rerun the test using a raw ZookeeperClient instead of a CuratorFramework?
On Thu, Nov 24, 2016 at 9:14 AM, Benjamin Jaton <[email protected]> wrote: > $ telnet localhost 60422 > Trying 127.0.0.1... > Connected to localhost. > Escape character is '^]'. > ruok > imokConnection closed by foreign host. > > > $ telnet localhost 60422 > Trying 127.0.0.1... > Connected to localhost. > Escape character is '^]'. > mntr > zk_version 3.5.2-alpha-1753710, built on 07/21/2016 16:24 GMT > zk_avg_latency 0 > zk_max_latency 0 > zk_min_latency 0 > zk_packets_received 5 > zk_packets_sent 3 > zk_num_alive_connections 1 > zk_outstanding_requests 0 > zk_server_state read-only > zk_znode_count 5 > zk_watch_count 0 > zk_ephemerals_count 0 > zk_approximate_data_size 230 > zk_open_file_descriptor_count 209 > zk_max_file_descriptor_count 4096 > Connection closed by foreign host. > > > Using zkCli.sh: > [zk: 127.0.0.1:60422(CONNECTEDREADONLY) 2] get /zookeeper/config > server.1=localhost:38328:57583:participant;localhost:60422 > server.2=localhost:47018:39326:participant;localhost:53253 > server.3=localhost:50001:58560:participant;localhost:39508 > version=0 > > > > > On Wed, Nov 23, 2016 at 1:49 PM, Cameron McKenzie <[email protected]> > wrote: > >> Can you connect via zkcli after you have shut down 2 of the servers in >> the cluster? Just trying to establish whether the cluster is responsive. >> >> On Thu, Nov 24, 2016 at 8:41 AM, Benjamin Jaton <[email protected] >> > wrote: >> >>> I'm sorry I am using Curator 3.2.1 (not 2.7.1) >>> >>> On Wed, Nov 23, 2016 at 1:39 PM, Benjamin Jaton < >>> [email protected]> wrote: >>> >>>> Sure, I'm using Curator 2.7.1 and ZK 3.5.2-alpha. >>>> I do set readonlymode.enabled in the code. >>>> >>>> Thanks >>>> Benjamin >>>> >>>> On Wed, Nov 23, 2016 at 1:15 PM, Vadim <[email protected]> wrote: >>>> >>>>> Benjamin, >>>>> >>>>> Can you specify Zookeeper version you use and whether you set up >>>>> "*readonlymode.enabled*" to true? By default read-only mode for ZK >>>>> server is disabled. >>>>> >>>>> Vadim. >>>>> >>>>> >>>>> >>>>> >>>>> On 2016-11-23 22:03, Benjamin Jaton wrote: >>>>> >>>>> Hello, >>>>> >>>>> Before I open a new JIRA for this, I wanted to bring it up here first. >>>>> >>>>> Basically the issue I'm seeing is that the read requests fail when >>>>> connecting to a read only ensemble. >>>>> >>>>> I made a test for this using TestingCluster, but it requires to set >>>>> iptables rules in order to mimic a real (remote) read-only ensemble. The >>>>> test will do this: >>>>> 1) start a TestingCluster >>>>> 2) stop 2 of the 3 nodes >>>>> 3) user has to run the iptables commands (specified in the logs) >>>>> 4) a read request is issued every 3 seconds >>>>> >>>>> All the read requests block for approx 2 minutes, then they all throw >>>>> an exception (ConnectionLoss). >>>>> >>>>> I was wondering if other users were using Curator with a read only >>>>> ensemble? >>>>> >>>>> Thanks >>>>> Benjamin >>>>> >>>>> >>>> >>> >> >
