This is a pre-release, handle with care!
This release contains breaking changes to both configuration and RPC protocol.
See https://zrepl.github.io/changelog.html for details.
Binaries were produced with the following docker container:
v0.1.0-rc4: digest: sha256:a522e5e0b2a0b9c8e605cc45e8d1ff27f660f6139fbc937c2b8e5e0072af76a7 size: 4103
Changes since 0.1.0-rc3
- The RPC protocol now uses gRPC for control RPCs and a custom protocol for data transfer
- The new implementation fixes the timeout problems we had with
go-streamrpc
- The new implementation fixes the timeout problems we had with
- The ssh transport now supports timeouts \o/
- We now require Go 1.11 or later
- The replication driver has been re-written to be more maintainable and handle the gRPC errors
- Status reporting changed a little
- Automatic retries on temporary network errors work for replication
- Automatic retries do not work for pruners (will not be in 0.1)
- The placeholder property now is a simple
on|offproperty- There is a migration subcommand, see changelog, please report any problems or unclear documentation!
- Thanks @JMoVS and @jbliesener for the discussion!
- The watchdog is gone, wasn't precise anyway and the new RPC timeout detection works quite reliably
- Snapshot & Pruning-only job type #101 (thanks @InsanePrawn )
- Configurable syslog facility (thanks @Ximalas )
- Systemd service defintion
- Grafana dashboard
- Bugfixes
- tls transport now supports intermediate CAs ( thanks @joshsouza )
- snapshotter hang-ups #152
- spaces in dataset names #131
- serialization of placeholder filesystem creation allows concurrent pushes to a single sink #153
- receive-side pruner now only prunes filesystems also present on the source-side
- Documentation of precautions for setups with > 2 jobs
Special thanks go to @JMoVS for many detailed bug reports!