2017-11-12 14:12:57 +01:00
.. |break_config| replace :: **[BREAK]**
2018-02-17 20:48:31 +01:00
.. |break| replace :: **[BREAK]**
2017-11-12 14:12:57 +01:00
.. |bugfix| replace :: [BUG]
2018-04-01 14:53:44 +02:00
.. |docs| replace :: [DOCS]
2018-02-17 20:48:31 +01:00
.. |feature| replace :: [FEATURE]
2017-11-12 14:12:57 +01:00
Changelog
=========
The changelog summarized bugfixes that are deemed relevant for users.
Developers should consult the git commit log or GitHub issue tracker.
2018-04-14 11:24:47 +02:00
0.0.4 (unreleased)
------------------
* |feature| :issue: `67` : Expose `Prometheus <https://prometheus.io> `_ metrics via HTTP (:ref: `config docs <monitoring-prometheus>` )
2018-02-17 20:48:31 +01:00
0.0.3
-----
* |break_config| |feature| :issue: `34` : automatic bookmarking of snapshots
* Snapshots are automatically bookmarked and pruning of bookmarks **must** be configured.
* This breaks existing configuration: `` grid `` :ref: `prune policy <prune-retention-grid>` specifications require the new `` keep_bookmarks `` parameter.
* Make sure to understand the meaning bookmarks have for :ref: `maximum replication downtime <replication-downtime>` .
* Example: :sampleconf: `pullbackup/productionhost.yml`
2018-03-04 17:19:41 +01:00
* |break| :commit: `ccd062e` : `` ssh+stdinserver `` transport: changed protocol requires daemon restart on both sides
* The delicate procedure of talking to the serving-side zrepl daemon via the stdinserver proxy command now has better error handling.
* This includes handshakes between client+proxy and client + remote daemo, which is not implemented in previous versions of zrepl.
* The connecting side will therefore time out, with the message `` dial_timeout of 10s exceeded `` .
* Both sides of a replication setup must be updated and restarted. Otherwise the connecting side will hang and not time out.
2018-02-18 13:35:40 +01:00
* |break_config| :commit: `2bfcfa5` : first outlet in `` global.logging `` is now used for logging meta-errors, for example problems encountered when writing to other outlets.
* |feature| :issue: `10` : `` zrepl control status `` subcommand
* Allows inspection of job activity per task and their log output at runtime.
* Supports `` --format raw `` option for JSON output, usable for monitoring from scripts.
* |feature| :commit: `d7f3fb9` : subcommand bash completions
* Package maintainers should install this as appropriate.
2018-02-18 13:28:46 +01:00
* |bugfix| :issue: `61` : fix excessive memory usage
2018-02-18 13:35:40 +01:00
* |bugfix| :issue: `8` and :issue: `56` : `` ssh+stdinserver `` transport properly reaps SSH child processes
* |bugfix| :commit: `cef63ac` : `` human `` format now prints non-string values correctly
* |bugfix| :issue: `26` : slow TCP outlets no longer block the daemon
2018-04-01 14:53:44 +02:00
* |docs| :issue: `64` : tutorial: document `` known_host `` file entry
2018-02-18 13:28:46 +01:00
2017-11-12 14:12:57 +01:00
0.0.2
-----
* |break_config| :commit: `b95260f` : `` global.logging `` is no longer a dictionary but a list
* |break_config| :commit: `3e647c1` : `` source `` job field `` datasets `` renamed to `` filesystems ``
* **NOTE** : zrepl will parse missing `` filesystems `` field as an empty filter,
i.e. no filesystems are presented to the other side.
* |bugfix| :commit: `72d2885` fix aliasing bug with root `<` subtree wildcard
* Filesystems paths with final match at blank `s` subtree wildcard are now appended to the target path
* Non-root subtree wildcards, e.g. `zroot/foo/bar<` still map directrly onto the target path
* Support days (`` d `` ) and weeks (`` w `` ) in durations
* Docs
* Ditch Hugo, move to Python Sphinx
* Improve & simplify tutorial (single SSH key per installation)
* Document pruning policies
* Document job types
* Document logging
* Start updating implementation overview
0.0.1
-----
* Initial release