Potboiler

Potboiler is an AP Event Sourcing system. More specifically, it's an MVP/research prototype of said, with known issues and is not even slightly suitable for production use. However, patches welcomed!

Getting started

1. pip install PyYAML
2. python generate-compose.py 3 > docker-compose.yml will generate node configs for 3 core/KV/browser sets
3. docker-compose up -d
4. Goto http://localhost:8002/8102/8202 to see any of the Key/Value browsing

Ports

The default Docker config gives you the following ports for the first node:

• Potboiler node: 8000
• Key/Value node: 8001
• Key/Value browser: 8002

For subsequent nodes, add 100 to the port numbers (e.g. 8100-2 for the second, 8200-2 for the third, etc)

API

Core

• List all current log heads

  • curl http://localhost:8000/log => {"69275a71-ec18-4be6-80a9-ac8e5d1d26b2":"d717f81d-dfc8-4c04-8fb3-1f28d63acf88"}

• List all log starts

  • curl http://localhost:8000/log/first => {"69275a71-ec18-4be6-80a9-ac8e5d1d26b2":"ad586c31-6cfe-43f2-9a7a-57e6371acff9"}

• Get log item

  • curl http://localhost:8000/log/6181ddc4-3c0b-4a40-b94c-f73379da886d => {"data":{"dfdsf":"sdfdsfs","foo":"bar"},"id":"6181ddc4-3c0b-4a40-b94c-f73379da886d","next":null,"owner":"69275a71-ec18-4be6-80a9-ac8e5d1d26b2","prev":"d717f81d-dfc8-4c04-8fb3-1f28d63acf88"}

• Add new log item

  • curl http://localhost:8000/log -d "{\"foo\":\"bar\", \"dfdsf\":\"sdfdsfs\"}" => redirect to "get log item"

• Register for log updates

  • curl http://localhost:8000/log/register -d "{\"url\": \"[URL to send msgs to]\"}" => 201

• Deregister for log updates

  • curl http://localhost:8000/log/deregister -d "{\"url\": \"[URL to send msgs to]\"}" => 204 if existed, otherwise 404

• List other nodes

  • curl http://localhost:8000/nodes => ["http://core1:8000","http://core0:8000"]

• Add new other node

  • curl http://localhost:8000/nodes -d "{\"url\": \"[Potboiler node root]\"}" => 201

KV

• Retrieve key
  • curl http://localhost:8001/kv/[table]/[key] => ["foo", "bar"]
  • 404 if no table, or no key
  • [table] and [key] here are alphanumeric strings

Update operations: "[item]" is any JSON value. "[key]" is a string.

• LWW

  • "set": "[item]"

• OR-Set:

  • "create": {} // makes an empty OR-Set if it doesn't exist
  • "add": {"key":"[key]", "item":"[item]", "metadata": "[metadata]"}
  • "remove": {"key":"[key]", "item":"[item]"}

• Update key

  • curl http://localhost:8001/kv/[table]/[key] -d "{\"op\": \"[operation]\", \"change\": \"[data]\"}" => Always 200 if data format is correct, regardless of whether the table has been seen

• Create table

  • Update key. "table" is "_config", "key" is table name. It's a LWW table, with "[item]" being {"crdt": "[crdt]"} "[crdt]" being only "LWW" current. Other info for the config table is ignored. e.g curl -vL http://localhost:8001/kv/_config/test -d "{\"op\": \"set\", \"change\": \"{\"crdt\": \"LWW\"}\"}"

Pigtail

Pigtail is a task queue implementation. Best docs for it at the moment are the example worker and provider in pigtail/example

Testing

• cargo test does most of it
• cargo test -- --ignored will also run the tests that need DATABASE_URL set