# What grpc address should I use to access the dgraph cluster

**URL:** <https://discuss.dgraph.io/t/what-grpc-address-should-i-use-to-access-the-dgraph-cluster/1898>\
**Category:** Users\
**Created:** [October 30, 2017, 12:37am UTC](https://discuss.dgraph.io/t/what-grpc-address-should-i-use-to-access-the-dgraph-cluster/1898 "2017-10-30T00:37:51Z")\
**Posts on this page:** 15\
**Page:** 1

<div class="post-metadata">

**Author:** ![jzhu077](https://avatars.discourse-cdn.com/v4/letter/j/da6949/32.png) [@jzhu077](https://discuss.dgraph.io/u/jzhu077)\
**Post date:** [October 30, 2017, 12:37am UTC](https://discuss.dgraph.io/t/what-grpc-address-should-i-use-to-access-the-dgraph-cluster/1898/1 "2017-10-30T00:37:51Z")

</div>

I set up a cluster consist of 3 dgraphzero instances + 3 dgraph nodes.  
dgraph are listening on (8080,9080), (8082,9082),(8084,9084)  
I notice if I use the web UI via localhost:8080 to ingest the nodes, the all 3 dgraph nodes are asking if it can server a particular predicate, but if I ingest via the grpc endpoint (ie. localhost:9080) I didn’t see other dgraph nodes apart from the one that is serving the 9080 port ask to serve the new predicate.

Setup

````auto
dgraphzero -w wz --bindall=true --my "localhost:8888"
dgraphzero -w wz1 --peer "localhost:8888" --port 8890 --bindall=true -idx 2
dgraphzero -w wz1 --peer "localhost:8888" --port 8892 --bindall=true -idx 3

dgraph --idx 1 --peer "localhost:8888" --my "localhost:12345" --bindall=true --memory_mb=2048
dgraph --idx 2 --peer "localhost:8888" --my "localhost:12346" --port 8082 --grpc_port 9082 --workerport 12346 --bindall=true --memory_mb=2048
dgraph --idx 3 --peer "localhost:8888" --my "localhost:12347" --port 8084 --grpc_port 9084 --workerport 12347 --bindall=true --memory_mb=2048```
````

---

<div class="post-metadata">

**Author:** ![peter](https://yyz1.discourse-cdn.com/flex007/user_avatar/discuss.dgraph.io/peter/32/574_2.png) [@peter](https://discuss.dgraph.io/u/peter)\
**Post date:** [October 30, 2017, 12:41am UTC](https://discuss.dgraph.io/t/what-grpc-address-should-i-use-to-access-the-dgraph-cluster/1898/2 "2017-10-30T00:41:57Z")

</div>

Hi @jzhu077, it doesn’t matter which grpc port you use for queries.

---

<div class="post-metadata">

**Author:** ![jzhu077](https://avatars.discourse-cdn.com/v4/letter/j/da6949/32.png) [@jzhu077](https://discuss.dgraph.io/u/jzhu077)\
**Post date:** [October 30, 2017, 12:46am UTC](https://discuss.dgraph.io/t/what-grpc-address-should-i-use-to-access-the-dgraph-cluster/1898/3 "2017-10-30T00:46:13Z")

</div>

hi @peter, it turns out all data are being written to the same node even though there are multiple predicates. I thought there should be some kind of parallel ingest and data sharding. It feels like I am running a single instance of dgraph instead of a cluster.

---

<div class="post-metadata">

**Author:** ![peter](https://yyz1.discourse-cdn.com/flex007/user_avatar/discuss.dgraph.io/peter/32/574_2.png) [@peter](https://discuss.dgraph.io/u/peter)\
**Post date:** [October 30, 2017, 12:49am UTC](https://discuss.dgraph.io/t/what-grpc-address-should-i-use-to-access-the-dgraph-cluster/1898/4 "2017-10-30T00:49:55Z")

</div>

There should definitely be data sharding occurring. The data for each predicate should be assigned to a different dgraph instance. Are you able to share the logs from the dgraph and dgraphzeros?

---

<div class="post-metadata">

**Author:** ![jzhu077](https://avatars.discourse-cdn.com/v4/letter/j/da6949/32.png) [@jzhu077](https://discuss.dgraph.io/u/jzhu077)\
**Post date:** [October 30, 2017, 1:04am UTC](https://discuss.dgraph.io/t/what-grpc-address-should-i-use-to-access-the-dgraph-cluster/1898/5 "2017-10-30T01:04:34Z")

</div>

Sorry about that, I made a mistake in the config after I reset the cluster. It’s working. Thanks for your help.

---

<div class="post-metadata">

**Author:** ![peter](https://yyz1.discourse-cdn.com/flex007/user_avatar/discuss.dgraph.io/peter/32/574_2.png) [@peter](https://discuss.dgraph.io/u/peter)\
**Post date:** [October 30, 2017, 1:08am UTC](https://discuss.dgraph.io/t/what-grpc-address-should-i-use-to-access-the-dgraph-cluster/1898/6 "2017-10-30T01:08:03Z")

</div>

All good, I’m glad you sorted it out.

---

<div class="post-metadata">

**Author:** ![jzhu077](https://avatars.discourse-cdn.com/v4/letter/j/da6949/32.png) [@jzhu077](https://discuss.dgraph.io/u/jzhu077)\
**Post date:** [October 30, 2017, 1:11am UTC](https://discuss.dgraph.io/t/what-grpc-address-should-i-use-to-access-the-dgraph-cluster/1898/7 "2017-10-30T01:11:34Z")

</div>

btw since I only have 3 dgraphzero instances, is there a point of adding more than 3 dgraph nodes? Does it help with improving the speed? I noticed when I create a dgraph node it logs

```auto
2017/10/30 13:53:35 draft.go:693: Restarting node for group: 3
2017/10/30 13:53:35 raft.go:567: INFO: 3 became follower at term 2
2017/10/30 13:53:35 raft.go:316: INFO: newRaft 3 [peers: [3], term: 2, commit: 17, applied: 17, lastindex: 17, lastterm: 2]
2017/10/30 13:53:39 raft.go:749: INFO: 3 is starting a new election at term 2
2017/10/30 13:53:39 raft.go:580: INFO: 3 became candidate at term 3
2017/10/30 13:53:39 raft.go:664: INFO: 3 received MsgVoteResp from 3 at term 3
2017/10/30 13:53:39 raft.go:621: INFO: 3 became leader at term 3
2017/10/30 13:53:39 node.go:301: INFO: raft.node: 3 elected leader 3 at term 3

```

Is it better to have multiple nodes serving the same group?

---

<div class="post-metadata">

**Author:** ![peter](https://yyz1.discourse-cdn.com/flex007/user_avatar/discuss.dgraph.io/peter/32/574_2.png) [@peter](https://discuss.dgraph.io/u/peter)\
**Post date:** [October 30, 2017, 1:21am UTC](https://discuss.dgraph.io/t/what-grpc-address-should-i-use-to-access-the-dgraph-cluster/1898/8 "2017-10-30T01:21:43Z")

</div>

In general, the number of `dgraphzero` instances can be different from the number of `dgraph` instances. We recommend having at least 3 `dgraphzero` instances.

Having more `dgraph` instances should spread the load so that each instance is doing less work, and therefore helps to increase throughput. You can also add replication as well, so that two (or more) dgraph instances are serving the same data.

There’s some documentation here in case you didn’t see it. [https://docs.dgraph.io/deploy/#multiple-instances](https://docs.dgraph.io/deploy/#multiple-instances)

---

<div class="post-metadata">

**Author:** ![jzhu077](https://avatars.discourse-cdn.com/v4/letter/j/da6949/32.png) [@jzhu077](https://discuss.dgraph.io/u/jzhu077)\
**Post date:** [October 30, 2017, 3:28am UTC](https://discuss.dgraph.io/t/what-grpc-address-should-i-use-to-access-the-dgraph-cluster/1898/9 "2017-10-30T03:28:27Z")

</div>

I am getting the single node ingesting when I am using the cluster. I thought I fixed it but I am still getting the single node ingesting… Looks like the first dgraph node is taking all of the predicates. Did I miss any thing in the setup?  
Here are the dgraphzero logs:

```auto
2017/10/30 16:19:38 node.go:554: RECEIVED: MsgAppResp 2-->1
2017/10/30 16:19:38 node.go:554: RECEIVED: MsgAppResp 2-->1
2017/10/30 16:19:38 node.go:554: RECEIVED: MsgAppResp 2-->1
2017/10/30 16:19:38 node.go:554: RECEIVED: MsgAppResp 2-->1
2017/10/30 16:19:38 node.go:554: RECEIVED: MsgAppResp 2-->1
2017/10/30 16:19:38 node.go:554: RECEIVED: MsgAppResp 2-->1
2017/10/30 16:19:38 node.go:554: RECEIVED: MsgAppResp 2-->1
2017/10/30 16:19:38 node.go:177: SENDING: MsgApp 1-->2
2017/10/30 16:19:38 node.go:554: RECEIVED: MsgReadIndex 2-->1
2017/10/30 16:19:38 node.go:554: RECEIVED: MsgAppResp 2-->1
2017/10/30 16:19:38 node.go:177: SENDING: MsgReadIndexResp 1-->2
2017/10/30 16:19:38 node.go:554: RECEIVED: MsgAppResp 2-->1
2017/10/30 16:19:38 node.go:554: RECEIVED: MsgAppResp 2-->1
2017/10/30 16:19:38 node.go:554: RECEIVED: MsgReadIndex 2-->1
2017/10/30 16:19:38 node.go:177: SENDING: MsgReadIndexResp 1-->2
2017/10/30 16:20:05 node.go:554: RECEIVED: MsgReadIndex 3-->1
2017/10/30 16:20:05 node.go:177: SENDING: MsgReadIndexResp 1-->3
2017/10/30 16:20:08 node.go:554: RECEIVED: MsgReadIndex 2-->1
2017/10/30 16:20:08 node.go:177: SENDING: MsgReadIndexResp 1-->2
2017/10/30 16:21:05 node.go:554: RECEIVED: MsgReadIndex 3-->1
2017/10/30 16:21:05 node.go:177: SENDING: MsgReadIndexResp 1-->3
2017/10/30 16:21:08 node.go:554: RECEIVED: MsgReadIndex 2-->1
2017/10/30 16:21:08 node.go:177: SENDING: MsgReadIndexResp 1-->2
2017/10/30 16:22:05 node.go:554: RECEIVED: MsgReadIndex 3-->1
2017/10/30 16:22:05 node.go:177: SENDING: MsgReadIndexResp 1-->3
2017/10/30 16:22:08 node.go:554: RECEIVED: MsgReadIndex 2-->1
2017/10/30 16:22:08 node.go:177: SENDING: MsgReadIndexResp 1-->2
2017/10/30 16:23:05 node.go:554: RECEIVED: MsgReadIndex 3-->1
2017/10/30 16:23:05 node.go:177: SENDING: MsgReadIndexResp 1-->3

```

```auto
2017/10/30 16:24:07 raft.go:253: Applied proposal: {Id:3895753382 Member:<nil> Tablet:group_id:1 predicate:"___commonkind" space:64762586 MaxLeaseId:0}
2017/10/30 16:24:07 node.go:177: SENDING: MsgAppResp 2-->1
2017/10/30 16:24:08 raft.go:253: Applied proposal: {Id:2289509746 Member:<nil> Tablet:group_id:1 predicate:"string_prop" space:57000000 MaxLeaseId:0}
2017/10/30 16:24:08 raft.go:253: Applied proposal: {Id:4155995118 Member:<nil> Tablet:group_id:1 predicate:"___orderable" space:55500000 MaxLeaseId:0}
2017/10/30 16:24:08 raft.go:253: Applied proposal: {Id:4271119824 Member:<nil> Tablet:group_id:1 predicate:"___child" space:43249800 MaxLeaseId:0}
2017/10/30 16:24:08 raft.go:253: Applied proposal: {Id:360075037 Member:<nil> Tablet:group_id:1 predicate:"int_prop" space:54500000 MaxLeaseId:0}
2017/10/30 16:24:08 raft.go:253: Applied proposal: {Id:113360965 Member:<nil> Tablet:group_id:1 predicate:"float_prop" space:55500000 MaxLeaseId:0}
2017/10/30 16:24:08 raft.go:253: Applied proposal: {Id:2763534684 Member:<nil> Tablet:group_id:1 predicate:"bool_prop" space:55500000 MaxLeaseId:0}
2017/10/30 16:24:08 raft.go:253: Applied proposal: {Id:3935270302 Member:<nil> Tablet:group_id:1 predicate:"_predicate_" space:173870115 MaxLeaseId:0}
2017/10/30 16:24:08 raft.go:253: Applied proposal: {Id:2931079240 Member:<nil> Tablet:group_id:1 predicate:"___kind" space:60762759 MaxLeaseId:0}
2017/10/30 16:24:08 node.go:177: SENDING: MsgAppResp 2-->1
2017/10/30 16:24:08 node.go:177: SENDING: MsgAppResp 2-->1
2017/10/30 16:24:08 node.go:177: SENDING: MsgAppResp 2-->1
2017/10/30 16:24:08 node.go:177: SENDING: MsgAppResp 2-->1
2017/10/30 16:24:08 node.go:177: SENDING: MsgReadIndex 2-->1
2017/10/30 16:24:08 node.go:554: RECEIVED: MsgReadIndexResp 1-->2
2017/10/30 16:24:08 node.go:177: SENDING: MsgReadIndex 2-->1
2017/10/30 16:24:08 node.go:554: RECEIVED: MsgReadIndexResp 1-->2
2017/10/30 16:24:08 node.go:177: SENDING: MsgReadIndex 2-->1
2017/10/30 16:24:08 node.go:554: RECEIVED: MsgReadIndexResp 1-->2

```

```auto
2017/10/30 16:24:07 raft.go:253: Applied proposal: {Id:3895753382 Member:<nil> Tablet:group_id:1 predicate:"___commonkind" space:64762586 MaxLeaseId:0}
2017/10/30 16:24:07 node.go:177: SENDING: MsgAppResp 3-->1
2017/10/30 16:24:08 raft.go:253: Applied proposal: {Id:2289509746 Member:<nil> Tablet:group_id:1 predicate:"string_prop" space:57000000 MaxLeaseId:0}
2017/10/30 16:24:08 raft.go:253: Applied proposal: {Id:4155995118 Member:<nil> Tablet:group_id:1 predicate:"___orderable" space:55500000 MaxLeaseId:0}
2017/10/30 16:24:08 raft.go:253: Applied proposal: {Id:4271119824 Member:<nil> Tablet:group_id:1 predicate:"___child" space:43249800 MaxLeaseId:0}
2017/10/30 16:24:08 raft.go:253: Applied proposal: {Id:360075037 Member:<nil> Tablet:group_id:1 predicate:"int_prop" space:54500000 MaxLeaseId:0}
2017/10/30 16:24:08 raft.go:253: Applied proposal: {Id:113360965 Member:<nil> Tablet:group_id:1 predicate:"float_prop" space:55500000 MaxLeaseId:0}
2017/10/30 16:24:08 raft.go:253: Applied proposal: {Id:2763534684 Member:<nil> Tablet:group_id:1 predicate:"bool_prop" space:55500000 MaxLeaseId:0}
2017/10/30 16:24:08 raft.go:253: Applied proposal: {Id:3935270302 Member:<nil> Tablet:group_id:1 predicate:"_predicate_" space:173870115 MaxLeaseId:0}
2017/10/30 16:24:08 raft.go:253: Applied proposal: {Id:2931079240 Member:<nil> Tablet:group_id:1 predicate:"___kind" space:60762759 MaxLeaseId:0}
2017/10/30 16:24:08 node.go:177: SENDING: MsgAppResp 3-->1
2017/10/30 16:24:08 node.go:177: SENDING: MsgAppResp 3-->1
2017/10/30 16:24:08 node.go:177: SENDING: MsgAppResp 3-->1
2017/10/30 16:24:08 node.go:177: SENDING: MsgAppResp 3-->1
2017/10/30 16:24:08 node.go:177: SENDING: MsgReadIndex 3-->1
2017/10/30 16:24:08 node.go:554: RECEIVED: MsgReadIndexResp 1-->3
2017/10/30 16:24:08 node.go:177: SENDING: MsgReadIndex 3-->1
2017/10/30 16:24:08 node.go:554: RECEIVED: MsgReadIndexResp 1-->3
2017/10/30 16:25:05 node.go:177: SENDING: MsgReadIndex 3-->1
2017/10/30 16:25:05 node.go:554: RECEIVED: MsgReadIndexResp 1-->3

```

dgraph:

```auto
Asking if I serve tablet: ___orderable
Asking if I serve tablet: string_prop
Asking if I serve tablet: ___orderable
Asking if I serve tablet: bool_prop
Asking if I serve tablet: ___orderable
Asking if I serve tablet: int_prop
Asking if I serve tablet: ___orderable
Asking if I serve tablet: bool_prop
Asking if I serve tablet: ___orderable
Asking if I serve tablet: string_prop
Asking if I serve tablet: ___orderable
Asking if I serve tablet: float_prop
Asking if I serve tablet: ___orderable
Asking if I serve tablet: string_prop
Asking if I serve tablet: ___orderable
Asking if I serve tablet: int_prop
Asking if I serve tablet: ___orderable
Asking if I serve tablet: float_prop
Asking if I serve tablet: ___orderable
Asking if I serve tablet: int_prop
Asking if I serve tablet: ___orderable
Asking if I serve tablet: float_prop
Asking if I serve tablet: ___orderable
Asking if I serve tablet: float_prop
Asking if I serve tablet: ___orderable
Asking if I serve tablet: float_prop
Asking if I serve tablet: ___orderable
Asking if I serve tablet: float_prop
Asking if I serve tablet: ___orderable
Asking if I serve tablet: float_prop
Asking if I serve tablet: ___orderable
Asking if I serve tablet: float_prop
Asking if I serve tablet: ___orderable
Asking if I serve tablet: float_prop
Asking if I serve tablet: ___orderable
Asking if I serve tablet: bool_prop
Asking if I serve tablet: ___orderable
Asking if I serve tablet: float_prop
Asking if I serve tablet: ___orderable
Asking if I serve tablet: float_prop
Asking if I serve tablet: ___orderable
Asking if I serve tablet: float_prop
Asking if I serve tablet: ___orderable
Asking if I serve tablet: float_prop
Asking if I serve tablet: ___orderable
Asking if I serve tablet: float_prop
Asking if I serve tablet: ___orderable
Asking if I serve tablet: bool_prop
Asking if I serve tablet: ___orderable
Asking if I serve tablet: float_prop
Asking if I serve tablet: ___orderable
Asking if I serve tablet: string_prop
Asking if I serve tablet: ___orderable
Asking if I serve tablet: bool_prop
Asking if I serve tablet: ___orderable
Asking if I serve tablet: string_prop
Asking if I serve tablet: ___orderable
Asking if I serve tablet: float_prop
Asking if I serve tablet: ___orderable
Asking if I serve tablet: float_prop
Asking if I serve tablet: ___orderable
Asking if I serve tablet: float_prop
Asking if I serve tablet: ___orderable
Asking if I serve tablet: float_prop
Asking if I serve tablet: ___orderable
Asking if I serve tablet: bool_prop
Asking if I serve tablet: ___orderable
Asking if I serve tablet: float_prop
Asking if I serve tablet: ___orderable
Asking if I serve tablet: string_prop
Asking if I serve tablet: ___orderable
Asking if I serve tablet: bool_prop
Asking if I serve tablet: ___orderable
Asking if I serve tablet: float_prop
Asking if I serve tablet: ___orderable
Asking if I serve tablet: float_prop
Asking if I serve tablet: ___orderable
Asking if I serve tablet: float_prop
Asking if I serve tablet: ___orderable
Asking if I serve tablet: float_prop
Asking if I serve tablet: ___orderable
Asking if I serve tablet: float_prop
Asking if I serve tablet: ___orderable
Asking if I serve tablet: int_prop
Asking if I serve tablet: ___orderable
Asking if I serve tablet: bool_prop
Asking if I serve tablet: ___orderable
Asking if I serve tablet: float_prop
Asking if I serve tablet: ___orderable
Asking if I serve tablet: bool_prop
Asking if I serve tablet: ___orderable
Asking if I serve tablet: float_prop
Asking if I serve tablet: ___orderable
Asking if I serve tablet: string_prop
Asking if I serve tablet: ___orderable
Asking if I serve tablet: float_prop
Asking if I serve tablet: ___orderable
Asking if I serve tablet: float_prop
Asking if I serve tablet: ___orderable
Asking if I serve tablet: int_prop
Asking if I serve tablet: bool_prop
Asking if I serve tablet: float_prop
Asking if I serve tablet: float_prop
Asking if I serve tablet: float_prop
Asking if I serve tablet: string_prop
Asking if I serve tablet: float_prop
Asking if I serve tablet: float_prop
Asking if I serve tablet: float_prop
Asking if I serve tablet: float_prop
Asking if I serve tablet: float_prop
Asking if I serve tablet: float_prop
Asking if I serve tablet: float_prop
2017/10/30 16:19:37 schema.go:90: Setting schema for attr string_prop: string, tokenizer: [], directive: NONE, count: false
2017/10/30 16:19:37 schema.go:90: Setting schema for attr ___orderable: bool, tokenizer: [], directive: NONE, count: false
2017/10/30 16:19:37 schema.go:90: Setting schema for attr float_prop: string, tokenizer: [], directive: NONE, count: false
2017/10/30 16:19:37 schema.go:90: Setting schema for attr int_prop: string, tokenizer: [], directive: NONE, count: false
2017/10/30 16:19:37 schema.go:90: Setting schema for attr bool_prop: string, tokenizer: [], directive: NONE, count: false
2017/10/30 16:21:05 wal.go:84: Writing snapshot to WAL: {Data:[9 1 0 0 0 0 0 0 0 16 1 26 15 108 111 99 97 108 104 111 115 116 58 49 50 51 52 53] Metadata:{ConfState:{Nodes:[1] XXX_unrecognized:[]} Index:25297 Term:2 XXX_unrecognized:[]} XXX_unrecognized:[]}
2017/10/30 16:22:05 wal.go:84: Writing snapshot to WAL: {Data:[9 1 0 0 0 0 0 0 0 16 1 26 15 108 111 99 97 108 104 111 115 116 58 49 50 51 52 53] Metadata:{ConfState:{Nodes:[1] XXX_unrecognized:[]} Index:36646 Term:2 XXX_unrecognized:[]} XXX_unrecognized:[]}
2017/10/30 16:23:05 wal.go:84: Writing snapshot to WAL: {Data:[9 1 0 0 0 0 0 0 0 16 1 26 15 108 111 99 97 108 104 111 115 116 58 49 50 51 52 53] Metadata:{ConfState:{Nodes:[1] XXX_unrecognized:[]} Index:39033 Term:2 XXX_unrecognized:[]} XXX_unrecognized:[]}
2017/10/30 16:25:05 wal.go:84: Writing snapshot to WAL: {Data:[9 1 0 0 0 0 0 0 0 16 1 26 15 108 111 99 97 108 104 111 115 116 58 49 50 51 52 53] Metadata:{ConfState:{Nodes:[1] XXX_unrecognized:[]} Index:39035 Term:2 XXX_unrecognized:[]} XXX_unrecognized:[]}

```

```auto
2017/10/30 16:19:07 HTTP server started. Listening on port 8082
2017/10/30 16:19:07 pool.go:104: == CONNECT ==> Setting localhost:8888
2017/10/30 16:19:08 groups.go:114: Connected to group zero. State: counter:19 groups:<key:1 value:<members:<key:1 value:<id:1 group_id:1 addr:"localhost:12345" leader:true last_update:1509333545 > > > > groups:<key:2 value:<members:<key:2 value:<id:2 group_id:2 addr:"localhost:12346" > > > > zeros:<key:1 value:<id:1 addr:"localhost:8888" leader:true > > zeros:<key:2 value:<id:2 addr:"localhost:8890" > > zeros:<key:3 value:<id:3 addr:"localhost:8892" > > 
2017/10/30 16:19:08 draft.go:143: Node ID: 2 with GroupID: 2
2017/10/30 16:19:08 pool.go:104: == CONNECT ==> Setting localhost:8890
2017/10/30 16:19:08 pool.go:104: == CONNECT ==> Setting localhost:8892
2017/10/30 16:19:08 pool.go:104: == CONNECT ==> Setting localhost:12345
2017/10/30 16:19:08 node.go:246: Group 2 found 0 entries
2017/10/30 16:19:08 draft.go:702: New Node for group: 2
2017/10/30 16:19:08 raft.go:567: INFO: 2 became follower at term 0
2017/10/30 16:19:08 raft.go:316: INFO: newRaft 2 [peers: [], term: 0, commit: 0, applied: 0, lastindex: 0, lastterm: 0]
2017/10/30 16:19:08 raft.go:567: INFO: 2 became follower at term 1
2017/10/30 16:19:08 groups.go:408: Unable to sync memberships. Error: rpc error: code = Unknown desc = Unknown cluster member
2017/10/30 16:19:08 raft.go:749: INFO: 2 is starting a new election at term 1
2017/10/30 16:19:08 raft.go:580: INFO: 2 became candidate at term 2
2017/10/30 16:19:08 raft.go:664: INFO: 2 received MsgVoteResp from 2 at term 2
2017/10/30 16:19:08 raft.go:621: INFO: 2 became leader at term 2
2017/10/30 16:19:08 node.go:301: INFO: raft.node: 2 elected leader 2 at term 2
2017/10/30 16:19:10 pool.go:104: == CONNECT ==> Setting localhost:12347

```

```auto
2017/10/30 16:19:09 worker.go:105: Worker listening at address: [::]:12347
2017/10/30 16:19:09 gRPC server started. Listening on port 9084
2017/10/30 16:19:09 pool.go:104: == CONNECT ==> Setting localhost:8888
2017/10/30 16:19:09 HTTP server started. Listening on port 8084
2017/10/30 16:19:10 groups.go:114: Connected to group zero. State: counter:21 groups:<key:1 value:<members:<key:1 value:<id:1 group_id:1 addr:"localhost:12345" leader:true last_update:1509333545 > > > > groups:<key:2 value:<members:<key:2 value:<id:2 group_id:2 addr:"localhost:12346" leader:true last_update:1509333548 > > > > groups:<key:3 value:<members:<key:3 value:<id:3 group_id:3 addr:"localhost:12347" > > > > zeros:<key:1 value:<id:1 addr:"localhost:8888" leader:true > > zeros:<key:2 value:<id:2 addr:"localhost:8890" > > zeros:<key:3 value:<id:3 addr:"localhost:8892" > > 
2017/10/30 16:19:10 draft.go:143: Node ID: 3 with GroupID: 3
2017/10/30 16:19:10 pool.go:104: == CONNECT ==> Setting localhost:8892
2017/10/30 16:19:10 pool.go:104: == CONNECT ==> Setting localhost:12345
2017/10/30 16:19:10 pool.go:104: == CONNECT ==> Setting localhost:12346
2017/10/30 16:19:10 pool.go:104: == CONNECT ==> Setting localhost:8890
2017/10/30 16:19:10 node.go:246: Group 3 found 0 entries
2017/10/30 16:19:10 draft.go:702: New Node for group: 3
2017/10/30 16:19:10 raft.go:567: INFO: 3 became follower at term 0
2017/10/30 16:19:10 raft.go:316: INFO: newRaft 3 [peers: [], term: 0, commit: 0, applied: 0, lastindex: 0, lastterm: 0]
2017/10/30 16:19:10 raft.go:567: INFO: 3 became follower at term 1
2017/10/30 16:19:10 groups.go:408: Unable to sync memberships. Error: rpc error: code = Unknown desc = Unknown cluster member
2017/10/30 16:19:10 raft.go:749: INFO: 3 is starting a new election at term 1
2017/10/30 16:19:10 raft.go:580: INFO: 3 became candidate at term 2
2017/10/30 16:19:10 raft.go:664: INFO: 3 received MsgVoteResp from 3 at term 2
2017/10/30 16:19:10 raft.go:621: INFO: 3 became leader at term 2
2017/10/30 16:19:10 node.go:301: INFO: raft.node: 3 elected leader 3 at term 2

```

disk size:

```auto
735M ./d1/p
28K ./d1/wz
278M ./d1/w
1012M ./d1
12K ./d2/p
28K ./d2/wz1
24K ./d2/w
68K ./d2
12K ./d3/p
28K ./d3/wz1
24K ./d3/w
68K ./d3
1013M ./

```

---

<div class="post-metadata">

**Author:** ![peter](https://yyz1.discourse-cdn.com/flex007/user_avatar/discuss.dgraph.io/peter/32/574_2.png) [@peter](https://discuss.dgraph.io/u/peter)\
**Post date:** [October 30, 2017, 3:34am UTC](https://discuss.dgraph.io/t/what-grpc-address-should-i-use-to-access-the-dgraph-cluster/1898/10 "2017-10-30T03:34:49Z")

</div>

What is the timeline of bringing up the second two dgraph instances? It can take some time for predicates to be transferred between instances. Predicate balance/transfer is only considered around once per 10 min.

How did you load the initial dataset? Via live loader, bulk loader, or some other solution?

---

<div class="post-metadata">

**Author:** ![jzhu077](https://avatars.discourse-cdn.com/v4/letter/j/da6949/32.png) [@jzhu077](https://discuss.dgraph.io/u/jzhu077)\
**Post date:** [October 30, 2017, 3:47am UTC](https://discuss.dgraph.io/t/what-grpc-address-should-i-use-to-access-the-dgraph-cluster/1898/11 "2017-10-30T03:47:51Z")

</div>

I started 3 dgraph intances up one after another. Then I was using a script to ingested the data via grpc. Does `Predicate balance/transfer is only considered around once per 10 min.` mean I should wait for 10 mins after I set up the schema?

---

<div class="post-metadata">

**Author:** ![mrjn](https://yyz1.discourse-cdn.com/flex007/user_avatar/discuss.dgraph.io/mrjn/32/775_2.png) [@mrjn](https://discuss.dgraph.io/u/mrjn)\
**Post date:** [October 30, 2017, 4:02am UTC](https://discuss.dgraph.io/t/what-grpc-address-should-i-use-to-access-the-dgraph-cluster/1898/12 "2017-10-30T04:02:24Z")

</div>

Yeah, currently just one move per 10 mins. If you want to evenly spread them out during ingestion, you could give the address of all the Dgraph servers to the client. It would then randomly pick up a dgraph server to send data to. That way, all servers would see new predicates and get to serve them.

For various reasons, if Dgraph server X is the first one to see a new predicate, it would be the one to serve it. That’s why doing this at the client level helps.

---

<div class="post-metadata">

**Author:** ![jzhu077](https://avatars.discourse-cdn.com/v4/letter/j/da6949/32.png) [@jzhu077](https://discuss.dgraph.io/u/jzhu077)\
**Post date:** [October 30, 2017, 4:29am UTC](https://discuss.dgraph.io/t/what-grpc-address-should-i-use-to-access-the-dgraph-cluster/1898/13 "2017-10-30T04:29:02Z")

</div>

thanks a lot for the suggestion. It would be nice if the 10mins predicate balancing and the suggestion is clearly stated in the documentation 🙂

---

<div class="post-metadata">

**Author:** ![mrjn](https://yyz1.discourse-cdn.com/flex007/user_avatar/discuss.dgraph.io/mrjn/32/775_2.png) [@mrjn](https://discuss.dgraph.io/u/mrjn)\
**Post date:** [October 30, 2017, 4:42am UTC](https://discuss.dgraph.io/t/what-grpc-address-should-i-use-to-access-the-dgraph-cluster/1898/14 "2017-10-30T04:42:38Z")

</div>

Please file a Github issue for the docs, we’ll fix the documentation.

---

<div class="post-metadata">

**Author:** ![system](https://canada1.discourse-cdn.com/flex007/uploads/dgraph/original/2X/2/2b38fdece2abe5814f2e51bee69d1e67fc304b0a.png) [@system](https://discuss.dgraph.io/u/system)\
**Post date:** [November 29, 2017, 4:57am UTC](https://discuss.dgraph.io/t/what-grpc-address-should-i-use-to-access-the-dgraph-cluster/1898/15 "2017-11-29T04:57:42Z")

</div>

This topic was automatically closed 30 days after the last reply. New replies are no longer allowed.