Tags

Release 2.7.0

* Major features

- switchover resign command was added. Now SRE can switch out
  RW-instances to available RO-instances by setting --selector rule

- switchover etcd load gets 2 new flags --append-only and --delete

- autofailover gets new option with_vshard_fix which fixes
  vshard.storage misconfiguration during autoswitch

- switchover status now implements option `-o json` which allows to
  use switchover in automation shell scripts

* Minor features

- switchover now implements internal DNS cache (TTL=60s) for resolve of
  tarantools listen-addr to ipv4 and to fqdn

- switchover pr/rr/eval gets 2 more flags --interval (as sleep interval)
  and --jobs (as parallel workers) to allow parallel execution

- switchover default log level now is WARN (was INFO)

* Breaking change

- switchover flag -t for commands switch/etcd wait/auto now can't be
  used. switchover -t is an alias for --etcd-timeout and used globally
  for all commands. To specify timeout of operation flag --timeout now
  should be used

Read more in Changelog.md

Release 2.6.0

* 2.6.0 - 2023-03-23

** Added

- `switchover watch` now respects `/instances/<instance_name>/autofailover/` configuration. User may set priorities for autofailover or disable promote of specific instances setting `/instances/<instance_name>/autofailover/role = "witness"`.
- `switchover watch` now exports telemetry and alerts into graphite if `autofailover.graphite` is configured
- switchover now mesures RTT for each connection and can report it via `switchover watch` telemetry
- `switchover pr / rr / eva` now supports flag `-i / --interval` to sleep between subsequent calls. (Default value is 0s as it was before)

** Fixed

- `Panic` now does not exit in `switchover watch`
- info logs of switchover now uses `green` color instead of `blue` color

** Removed

- `switchover` now does not build for DEB

Release 2.5.2 - 2023-02-06

Release 2.5.1

* Added
- `switchover status` now shows vshard alerts and WAL alerts

* Fixed
- `switchover top` fixed search cursor position

Release 2.5.0

** Added

- `switchover etcd health` now requests /health endpoint from etcd
- `switchover etcd version` now requests /version endpoint from etcd
- `switchover etcd getr` added as an alias to `switchover etcd get -r`
- `switchover connect --ps1` flag added to modify tarantool console prompt (experimental)
- `switchover top` now runs in interactive mode
- `switchover --no-etcd` flag is added to disable etcd

** Fixed
- `switchover auto` now fences master before changing /<cluster>/master in ETCD
- `switchover --completion` fixed
- `switchover switch <shard>:ro` now supported to switch master to any replica
- `switchover promote <shard>:rw` now supported to promote to etcd defined master
- `rockspec_format` is changed to 1.1
- etcd version in tests was upgraded to 3.5.3
- switchover directories was changed: /usr/share/lua/5.1 -> /usr/share/tarantool
- `switchover top` replication lag was fixed

Release 2.4.0 - 2022-12-16

Release 2.3.2 - 2022-11-09

*** Added

- `switchover etcd validate` added to validate configuration for Tarantool application
- `switchover status` now shows tarantool version
- `switchover eval --shard <shard-name> <shell>` and
  `switchover eval --cluster <cluster-name> <shell>` was added.

*** Fixed

- `switchover {switch,promote}` and others now properly log
  after-switch discovery info (as it was before 2.3.0)

*** Removed

- `switchover eval --all` now removed.
  You should use `--shard` and/or `--cluster` instead.

2.3.1 - 2022-09-05

*** Added

- switchover top: now accepts input as `switchover top cluster_name/shard_name` and reduces output to statistics and alerts of shard (not entire cluster)
- switchover status: now accepts input as `switchover status cluster_name/shard_name` and reduces output to statistics and alerts of shard (not entire cluster)

*** Fixed

- Fixed Background fibers now closed explicitely using background.registry for `package-reload`, `restart-replication` and `switch` commands. This patch removes annoying logs.

- switchover top: added `Conns` column which stands for `Established connections to tarantool (taken from box.stat.net().CONNECTIONS.current`
- switchover top: `Cur` was renamed to `Reqs` (stands for network requests per second).

- github.com/ochaton/background was imported to switchover. All meaningfull background fibers were rewritten using background library
- switchover restart-replication (rr) now behaves in the same way as package-reload (pr). rr also supports --cluster and --shard flags.

- Fixed fiber leaks and eval flood during `switchover watch`.
- Sockets to unconnected tarantools now closed after discovery timeout exhausted.

- Nothing was removed

- No security changes were introdusced.

Switchover 2.2 (pre-relase)

    * (alpha) switchover top
    * switchover --ask-pass and work with credentials
    * switchover promote now drop checks for downstrems and
      upstreams
    * switch now can reload routers if etcd.cluster.vshard ETCD
      config is used
    * switchover auto can choose best candidate to promote
    * switchover etcd load now supports json and yaml for input and
      output
    * switchover eval now works with shard
    * switchover now can switch Cartridge Cluster if Statefull
      Failover is enabled
    * switchover now can load cluster topology if vshard or cluster
      have met
    * switchover discovery now accepts many discovery-formats
    * switchover now can print it's version
    * switchover now can show status for proxies
      (etcd.instance.single)
    * switchover now can load config from $HOME/.switchover.yaml

Switchover 2.1

	* switchover status - status of the cluster
	* switchover eval - run any command on the shard/instance
	* switchover --discovery-format - set any format

Switchover 2.0

	* Upgrades interface of cli. Option `--cluster` is removed
	  instead used human friendly notation: mycrud@mycrud_001
	* Fixed behaviour of `--discovery-timeout` in all commands.
	* Extended logs with date, time and colors
	* Support of `switchover etcd` allows to get information from
	  ETCD and fix some values in it
	* Support of `switchover connect` allows to run console to
	  specified instance. Ex. `switchover connect mycrud:master`
	* Added flags to promote instance during Emergency. Such as
	  `--no-check-*`, `--valid-status` and `--no-fence-master`.
	* `switchover promote` now tries to switch to RO old broken
	  master. Helpfull during network split.

Release 1.0