Open yuandongjian opened 3 years ago
@crazy-pizza that error comes up if the coordinator cannot connect to m3db. Are you able to confirm it's able to connect?
Also the etcd cluster endpoints you have listed seem potentially suspect, are you services connecting healthily to the underlying etcd cluster?
@wesleyk It seems that coordinator can connect to m3DB normally. After namespace is set to ready with force params, the cluster can also read and write data normally. Everything seems to be working just fine, except for the need to force the namespace.
It seems that m3DB has connected to the ETCD cluster, and there is no error in the M3DB log, what's wrong with etcd cluster endpoints?
Here is the data in ectd
[root@ etcd]# ./etcdctl --endpoints=$ENDPOINTS get --prefix "" --keys-only=true
/placement/default_env/m3aggregator
/placement/namespace/m3db-cluster-name/m3aggregator
/topic/namespace/m3db-cluster-name/aggregated_metrics
/topic/namespace/m3db-cluster-name/aggregated_metrics2
/topic/namespace/m3db-cluster-name/aggregator_ingest
_kv/default_env/m3db.node.namespaces
_kv/namespace/m3db-cluster-name/shardset/1/flush
_sd.placement/default_env/m3db
_sd.placement/namespace/m3db-cluster-name/m3coordinator
@crazy-pizza sorry for the delay. At this point it'll be harder to debug, though if you're able to reproduce from fresh, then providing coordinator and db logs would be helpful.
M3DB unable to mark namespace as ready without force params.
m3db version: 1.1.0
m3coordinator.yml
m3dbnode.yml