swiftbunny

CLI/daemon to help with log archiving to OpenStack Swift


Keywords
archiving, openstack-swift, rabbitmq
License
GPL-3.0+
Install
pip install swiftbunny==1.1

Documentation

SwiftBunny :: CLI/daemon to help with log archiving to OpenStack Swift

Reads RabbitMQ, reads Graylog+ElasticSearch JSON dumps.

Writes daily/hourly logs files to OpenStack Swift.

Encrypts data on the fly using qpress (.qz1) and GnuPG.

Why? Organised long term storage of logs.

Output file format

Time is duplicated in the file, as the datetime is taken from the JSON object. The compression will take care of excess size:

2020-01-25T00:00:12.217000Z: {"some": "json", "dict": "..."}
2020-01-25T00:00:12.496000Z: {"another": "json", "object": "..."}
...

Target file names will look like:

example-swift-container-2020/example.rabbit.queue/2020/05/31/
  example.rabbit-queue_2020-05-25T00:00:00Z+3600-0x72272_keyid.qz1.gpg

Which means:

  • the target container is example-swift-container-2020;
  • the source queue is example.rabbit.queue;
  • the first message is after 2020-05-25, at 00:00:00 (UTC);
  • the last message is before that time +3600 (i.e. one hour later);
  • there are no late message (-0);
  • the file contains 72272 lines (records).
  • keyid may describe which node this was produced on, and a short identifier which symmetric gpg key was used.

So, if you're looking for old records, you can:

  • find all files in the right period (also check nearby files with -N-times);
  • fetch, decrypt and decompress the files;
  • scan the lines by the consistently formatted timestamps.

TODO

  • Add commands, so we can do 'ls' and 'get' and 'post' on the archived files. (Use argparse.)
  • Add sample ini-file.
  • Fix suffix (split node-X and keyname).
  • Add Immutable middleware to Swift so we can make it harder to accidentally delete stored logs.
  • Don't forget about chmod 600 in swiftbunny.ini and chmod 700 on /var/spool/swiftbunny.