Hi, So, is the case reproduced only when the nodes are started simultaneously? Could you add any delay between instances starts?
Looks like an issue. Please provide the logs, please. On Thu, Dec 1, 2016 at 1:29 AM, ghughal <bakul.ghug...@gmail.com> wrote: > We are seeing intermittent issue where ignite node is not able to join the > cluster for our specific configuration. Our ignite instances are part of > spring boot application and it’s running inside docker container. These > docker container gets deployed by marathon framework on mesos. When we try > to start 2 instances, marathon starts both instance at about same time and > it seems Ignite is getting into some race condition. > > Since we are running inside docker container, we are using > TcpDiscoveryZookeeperIpFinder with AddressResolver. Here’s code for that: > > > > It seems each node starts at same time and checks with zookeeper if there > are any other nodes in cluster. It doesn’t find any (in most cases) and > creates separate cluster. If we destroy one of the instances and start it > again it joins the cluster and everything works fine. > > > > > -- > View this message in context: http://apache-ignite-users. > 70518.x6.nabble.com/Not-able-to-join-cluster-with- > Zookeeper-based-IP-finder-tp9311.html > Sent from the Apache Ignite Users mailing list archive at Nabble.com. >
