These are chat archives for atomix/atomix

18th
Jan 2018
Bruce Grobler
@zabda_twitter
Jan 18 2018 07:37
Morning all (well in my timezone), i have a three node cluster, which provides a simple consistent map ("foobar"), all works fine when the three nodes are up and running,
however if i turn off one of them, its not happy at all,
atomix-poc_2_1 | [2018-01-18 07:36:34.897] WARN [e-partition-1-7]
atomix-poc_2_1 | .i.PrimaryBackupServiceContext V-Trace-ID=[] - PrimitiveService{data-partition-45}{type=CONSISTENT_MAP, name=foobar} - Not enough backups; transitioning to NONE
atomix-poc_2_1 | Role: "NONE"
that happens when i query the map thereafter,
i have looked at the code and the builder defaults to two backups, if not specified,
am i misunderstanding something ?
[
    {
        "id": "10.5.0.33",
        "type": "DATA",
        "endpoint": "10.5.0.33:8700",
        "state": "ACTIVE"
    },
    {
        "id": "10.5.0.32",
        "type": "DATA",
        "endpoint": "10.5.0.32:8700",
        "state": "ACTIVE"
    },
    {
        "id": "10.5.0.31",
        "type": "DATA",
        "endpoint": "10.5.0.31:8700",
        "state": "INACTIVE"
    }
]
a new leader was elected after the failure,
atomix-poc_3_1        |                           i.a.p.r.impl.RaftContext V-Trace-ID=[] - RaftServer{coordination-partition-2} - Transitioning to LEADER
atomix-poc_3_1        | Role: "LEADER"
atomix-poc_3_1        | [2018-01-18 07:36:29.409] INFO  [ore-partition-1]
atomix-poc_3_1        |                           i.a.p.r.impl.RaftContext V-Trace-ID=[] - RaftServer{core-partition-1} - Found leader 10.5.0.33
atomix-poc_3_1        | NodeId: { }
Bruce Grobler
@zabda_twitter
Jan 18 2018 07:46
and if i bring the original failure node back online, it rejoins, but the map is unavailable
atomix-poc_2_1        |            c.d.v.m.RestControllerAdvice V-Trace-ID=[yFciiVuQkwenlE2IyKpnfy] - foobar
atomix-poc_2_1        |   io.atomix.core.map.ConsistentMapException$Timeout: foobar
atomix-poc_2_1        |         at io.atomix.core.map.impl.BlockingConsistentMap.complete(BlockingConsistentMap.java:237)
atomix-poc_2_1        |         at io.atomix.core.map.impl.BlockingConsistentMap.addListener(BlockingConsistentMap.java:187)
atomix-poc_2_1        |         at io.atomix.core.map.ConsistentMap.addListener(ConsistentMap.java:344)
atomix-poc_2_1        |         at com.devshackintl.rnd.atomix.cluster.AtomixManager.getMap(AtomixManager.java:41)
any ideas where i might be going wrong ?
Johno Crawford
@johnou
Jan 18 2018 11:08
@zabda_twitter each node has its own data directory, correct?
Bruce Grobler
@zabda_twitter
Jan 18 2018 12:16
ye it does
or rather, they do,