Setup DGraph Cluster on ( Ubuntu20.04, Jammy) - 3 Alpha and 3 Zero ( HA, Recovery, Snapshot, Visibility )

Ashwani_DU-MCA · July 2, 2024, 6:20am

What I want to do

Setup cluster of DGraph in Ubuntu20.04 with 3 Alpha and 3 Zero

What I did

Steps to run as daemon:

dpkg --print-architecture
mkdir dgraph, cd dgraph
wget https://github.com/dgraph-io/dgraph/releases/download/v23.1.0/dgraph-linux-amd64.tar.gz
tar -xvzf dgraph-linux-amd64.tar.gz
create file for daemon: vi /etc/systemd/system/dgraph.service

[Unit]
Description=dgraph.io data server
Wants=network.target
After=network.target dgraph.service
Requires=dgraph.service

[Service]
Type=simple
ExecStart=/home/sre/dgraph/dgraph alpha --my=VM_IP:7080 --zero=VM_IP:5080 --raft=idx=2 -p /var/run/dgraph/p -w /var/run/dgraph/w
StandardOutput=journal
StandardError=journal
User=root
Group=root

[Install]
WantedBy=multi-user.target

mkdir -p /var/run/dgraph/zw, chmod -R 750 /var/run/dgraph
ls -ld /var/run/dgraph, ls -ld /var/run/dgraph/zw/
sudo systemctl daemon-reload, systemctl status dgraph, systemctl restart dgraph

Post changing the above configuration for different VM’s:

Able to run the DGraph with 3 Alpha and 3 Zero Nodes as Daemon Service in Ubuntu20.04

Zero 1:
ExecStart=/home/sre/dgraph/dgraph zero --my=VM_IP_1:5080 --replicas 3 --raft=idx=1 --wal /var/run/dgraph/zw

Zero 2: 
ExecStart=/home/sre/dgraph/dgraph zero --my=VM_IP_2:5080 --peer VM_IP_1:5080 --replicas 3 --raft=idx=2 --wal /var/run/dgraph/zw

Zero 3:
ExecStart=/home/sre/dgraph/dgraph zero --my=VM_IP_3:5080 --peer VM_IP_1:5080 --replicas 3 --raft=idx=3 --wal /var/run/dgraph/zw


Alpha 1:
ExecStart=/home/ashwani.yadav/dgraph alpha --my=VM_IP_4:7080 --zero=VM_IP_1:5080 --raft=idx=1 -p /var/run/dgraph/p -w /var/run/dgraph/w

Alpha 2:
ExecStart=/home/sre/dgraph/dgraph alpha --my=VM_IP_5:7080 --zero=VM_IP_1:5080 --raft=idx=2 -p /var/run/dgraph/p -w /var/run/dgraph/w

Alpha 3:
ExecStart=/home/sre/dgraph/dgraph alpha --my=VM_IP_6:7080 --zero=VM_IP_1:5080 --raft=idx=3 -p /var/run/dgraph/p -w /var/run/dgraph/w

Now cluster is up and running without any issue, but in the RatelUI only 1 Group is having all the tablets, none of the other Group of Alpha has any tablets, until and unless I am specifically making the schema in it.

My questions are:

Ideally Tablets should be distributed across different Alpha’s in the Group automatically, right ?

Second Issue, I am not able to get my data back. Insertion of data went fine and have given me success on the admin API response, but once I am trying to find the data using Query nothing is showing up ?

How to maintain the snapshot and how to apply snapshot in case recovery is needed ?

Ashwani_DU-MCA · July 2, 2024, 6:22am

Other Images for Query, Predicate, Types visible

Ashwani_DU-MCA · July 2, 2024, 7:50am

schema:

curl -X POST localhost:8080/alter -d '
type Customer {
  id
  name
  email
  purchase_history
}
'

Logs:

Jul 02 13:14:17 stg-ashwani005 dgraph[2569560]: I0702 13:14:17.043551 2569588 http.go:614] Got alter request via HTTP from 127.0.0.1:22784
Jul 02 13:14:17 stg-ashwani005 dgraph[2569560]: I0702 13:14:17.044506 2569588 server.go:377] Received ALTER op: schema:"\nid: string @index(exact) @upsert .\nname: string @index(term) .\nemail: string @index(exact) .\npurchase_history: [uid] .\n"
Jul 02 13:14:17 stg-ashwani005 dgraph[2569560]: I0702 13:14:17.044547 2569588 server.go:1686] Got Alter request from: "127.0.0.1:22784"
Jul 02 13:14:17 stg-ashwani005 dgraph[2569560]: I0702 13:14:17.046324 2569588 server.go:526] Got schema: &{Preds:[predicate:"0-id" value_type:STRING directive:INDEX tokenizer:"exact" upsert:true  predicate:"0-name" value_type:STRING directive:INDEX tokenizer:"term"  predicate:"0-email" value_type:STRING directive:INDEX tokenizer:"exact"  predicate:"0-purchase_history" value_type:UID list:true ] Types:[]}
Jul 02 13:14:17 stg-ashwani005 dgraph[2569560]: I0702 13:14:17.047724 2569588 mutation.go:202] Max open files limit: 524288
Jul 02 13:14:17 stg-ashwani005 dgraph[2569560]: I0702 13:14:17.048770 2569588 index.go:780] Deleting indexes for 0-id
Jul 02 13:14:17 stg-ashwani005 dgraph[2569560]: I0702 13:14:17.049084 2569588 index.go:780] Deleting indexes for 0-name
Jul 02 13:14:17 stg-ashwani005 dgraph[2569560]: I0702 13:14:17.050666 2569588 server.go:540] ALTER op: schema:"\nid: string @index(exact) @upsert .\nname: string @index(term) .\nemail: string @index(exact) .\npurchase_history: [uid] .\n"  done
Jul 02 13:14:17 stg-ashwani005 dgraph[2569560]: I0702 13:14:17.063793 2569588 http.go:614] Got alter request via HTTP from 127.0.0.1:22786
Jul 02 13:14:17 stg-ashwani005 dgraph[2569560]: I0702 13:14:17.064048 2569588 server.go:377] Received ALTER op: schema:"\ntype Customer {\n  id\n  name\n  email\n  purchase_history\n}\n"
Jul 02 13:14:17 stg-ashwani005 dgraph[2569560]: I0702 13:14:17.064088 2569588 server.go:1686] Got Alter request from: "127.0.0.1:22786"
Jul 02 13:14:17 stg-ashwani005 dgraph[2569560]: I0702 13:14:17.065496 2569588 server.go:526] Got schema: &{Preds:[] Types:[type_name:"0-Customer" fields:<predicate:"0-id" > fields:<predicate:"0-name" > fields:<predicate:"0-email" > fields:<predicate:"0-purchase_history" > ]}
Jul 02 13:14:17 stg-ashwani005 dgraph[2569560]: I0702 13:14:17.067948 2569588 server.go:540] ALTER op: schema:"\ntype Customer {\n  id\n  name\n  email\n  purchase_history\n}\n"  done

Insertion of data:

curl -X POST localhost:8080/mutate -H "Content-Type: application/json" -d '{
  "set": [
    {
      "uid": "_:customer1",
      "dgraph.type": "Customer",
      "id": "1",
      "name": "Alice Johnson",
      "email": "alice@example.com"
    },
    {
      "uid": "_:customer2",
      "dgraph.type": "Customer",
      "id": "2",
      "name": "Bob Smith",
      "email": "bob@example.com"
    }
  ]
}'

Response I get:

{"data":{"code":"Success","message":"Done","queries":null,"uids":{"dg.960315563.1":"0x2720","dg.960315563.2":"0x2723","purchase1":"0x2721","purchase2":"0x2722","purchase3":"0x2724","purchase4":"0x2725"}},"extensions":{"server_latency":{"parsing_ns":191650,"processing_ns":3704405,"assign_timestamp_ns":1295739,"total_ns":5490665},"txn":{"start_ts":10348,"keys":["11h35u3v567up","1ddilwjdppv2j","1gmlrx2gms5t0","1gmlrx2gms5t1","1olx0ehdj620m","1zwvmp6gd98tv","28c6ermowwj5r","2przdhzp04s6w","2q5p2hgu1umdc","2qgxm4hdfytlh","2r56n110w1fjv","2u3bihi2lt2vy","2uaugmpvcheqj","2vlq4maf743ss","32zg2v1id4btk","33g4qzj0afh38","33g4qzj0afh3b","3k4e946smtzr8","8p8gv9lzzj9y","9726oyo3iqva","faz4bt8juuyt","faz4bt8juuyu"],"preds":["1-0-dgraph.type","1-0-id","1-0-name","2-0-email","2-0-purchase_history"]}}}

Logs
Nothing got updated in the logs

Query I used to fetch the data:

curl -H "Content-Type: application/json" -X POST localhost:8080/query -d '{
  "query": "{ allCustomers(func: has(dgraph.type)) { uid dgraph.type id name email } }"
}'

Response:

{"data":{"allCustomers":[{"uid":"0x1","dgraph.type":["dgraph.graphql"]}]},"extensions":{"server_latency":{"parsing_ns":58848,"processing_ns":2382858,"encoding_ns":108988,"assign_timestamp_ns":1150921,"total_ns":3813301},"txn":{"start_ts":10360},"metrics":{"num_uids":{"_total":5,"dgraph.type":1,"email":1,"id":1,"name":1,"uid":1}}}}

Logs:
Nothing got updated in the logs

RatelUI Query:

{
  allCustomers(func: has(dgraph.type)) {
    uid
    dgraph.type
    id
    name
    email
  }
}

Response:

{
  "data": {
    "allCustomers": [
      {
        "uid": "0x1",
        "dgraph.type": [
          "dgraph.graphql"
        ]
      }
    ]
  },
  "extensions": {
    "server_latency": {
      "parsing_ns": 85408,
      "processing_ns": 1532104,
      "encoding_ns": 57038,
      "assign_timestamp_ns": 1324574,
      "total_ns": 3156200
    },
    "txn": {
      "start_ts": 10378
    },
    "metrics": {
      "num_uids": {
        "_total": 5,
        "dgraph.type": 1,
        "email": 1,
        "id": 1,
        "name": 1,
        "uid": 1
      }
    }
  }
}

Shivaji_Kharse · July 2, 2024, 8:45am

try this query

  allCustomers(func: type(Customer)) {
    uid
    dgraph.type
    id
    name
    email
  }
}

Ashwani_DU-MCA · July 2, 2024, 9:17am

able to setup the cluster properly it seems: Cluster Types - Deploy

updated the configuration above for Alpha, we have to make Zero as per my first thread, but in Alpha we don’t have to give any raft_id to create the group.

But still I am some queries in regard to Snapshot and Backup. How to achieve this as the Docs does not specify anything in brief for manual cluster management.

Topic		Replies	Views
Set up dgraph cluster manually Users	8	1469	June 1, 2019
Setting up cluster with docker-compose Dgraph kind:bug , docker	0	686	April 16, 2022
Dgraph db cluster in 3 data centers Dgraph	6	428	February 16, 2023
Running multiple Dgraph alpha pod and zeros in single host Dgraph kind:question	11	580	March 23, 2023
Clarification on machine requirements for Dgraph Dgraph kind:question , dgraph	1	588	May 2, 2023

Setup DGraph Cluster on ( Ubuntu20.04, Jammy) - 3 Alpha and 3 Zero ( HA, Recovery, Snapshot, Visibility )

What I want to do

What I did

Related Topics