Are snapshots working using Docker compose?

Kubernetes
1

I’ve been running Dragonflydb on a single node using Docker compose for about 8 months now. I am using the following docker compose yaml to attempt to take snapshots every 5 minutes:

# This version persists data.
services:
  dragonfly:
    image: 'docker.dragonflydb.io/dragonflydb/dragonfly'
    command: ["dragonfly", "--logtostderr", "--dir", "/data", "--snapshot_cron", "*/5 * * * *"] # Example cron schedule for every 5 minutes
    # Then you will also have to have a cron job to delete old snapshots
    ulimits:
      memlock: -1
    restart: unless-stopped
    #ports:
    #  - "6379:6379"
    # For better performance, consider `host` mode instead `port` to avoid dock>
    # `host` mode is NOT currently supported in Swarm Mode.
    # https://docs.docker.com/compose/compose-file/compose-file-v3/#network_mode
    network_mode: "host"
    volumes:
       - dragonflydata:/data
volumes:
  dragonflydata:

This code runs the db fine, but when I look at

/var/lib/docker/volumes/redis_dragonflydata/_data

Timestamps are created when I first start the container but never updated afterwards. If the snapshot files are getting updated then I would expect the filesystem timestamps to change.

Either something’s not working or I have some incorrect assumptions.

Thanks for any help.

2

Hi @windmeup, good question!

I think the gist is that you don’t pass dragonfly as the first argument. I tested it myself, and all the following 3 ways could work.

Command as a Whole String

services:
  dragonfly:
    image: "docker.dragonflydb.io/dragonflydb/dragonfly"
    command: "--proactor_threads 2 --snapshot_cron '*/5 * * * *'"
    ulimits:
      memlock: -1
    ports:
      - "6379:6379"
    restart: unless-stopped
    volumes:
      - dragonflydata:/data
volumes:
  dragonflydata:

Command as a YAML List

services:
  dragonfly:
    image: "docker.dragonflydb.io/dragonflydb/dragonfly"
    command:
      - "--proactor_threads"
      - "2"
      - "--snapshot_cron"
      - "*/5 * * * *"
    ulimits:
      memlock: -1
    ports:
      - "6379:6379"
    restart: unless-stopped
    volumes:
      - dragonflydata:/data
volumes:
  dragonflydata:

Command as a Compact List

This is probably the way you prefer:

services:
  dragonfly:
    image: "docker.dragonflydb.io/dragonflydb/dragonfly"
    command: ["--proactor_threads", "2", "--snapshot_cron", "*/5 * * * *"]
    ulimits:
      memlock: -1
    ports:
      - "6379:6379"
    restart: unless-stopped
    volumes:
      - dragonflydata:/data
volumes:
  dragonflydata:

After Dragonfly is booted, you can connect and verify the snapshot_cron configuration like below:

$> redis-cli # Note that I have the port accessible locally.

dragonfly$> CONFIG GET snapshot_cron
1) "snapshot_cron"
2) "*/5 * * * *"

I also monitored, seeing that snapshots are created based on the cron schedule:

image
image1043×548 72.2 KB

Thanks for the question. I will update the documentation with examples like this shortly.

3

You are a prince! :wink:

Thank you!

1 Like
4

Question; does that mean that I no longer need these?"

“–logtostderr”, “–dir”, “/data”,

6

You can still pass additional server flags as needed!

I was using "--proactor_threads", "2", "--snapshot_cron", "*/5 * * * *" earlier just as an example, as I was running it locally too and wanted to have a smaller memory requirement and test the flags.

7

I did pass some more and it’s working great. I also wrote a script to be run on cron every 15 minutes to leave just the last 4 days of logs.

1 Like