Release notes from fleet

v1.0.0

2016-12-22T14:52:35Z

Release from tag v1.0.0, commit b8127af ("Merge pull request #1719 from jonboulle/master"), as of 2016-12-22.

Note that this is the final anticipated release of fleet and no future activity is planned. CoreOS will no longer sponsorship any continued maintenance or development of the project. If you are interested in the future of the project and wish to take over ownership, please contact [email protected]">[email protected].

This release includes new features, especially gRPC communication, basic etcd authentication, and support for Install section. It also fixes numerous bugs that have been discovered recently. For details, please see the list below.

New Features

#1393: fleetd: Support for targeting "systemd --user"
#1426: Use gRPC to communicate the engine and agents
#1564: fleetd: get filtered list of units instead of all units
#1571: fleetd: support conflict in global unit
#1624: fleetctl: status and restart commands with global units
#1630: config: add basic etcd authentication
#1642: fleetd: support dynamic metadata (combined with #1703: api,functional: dynamic metadata - fix bugs and add functional tests)
#1655: fleetd: process dependencies in [Install] section
#1673: registry/rpc: use simpleBalancer instead of ClientConn.State()
#1679: api: support cAPI.UnitState() for a single unit
#1681: fleetd,main: make github.com/coreos/fleet go-getable
#1695: agent: allow HasConflict() to handle multiple values defined in Conflicts

Bugs fixed

#1503: fleetctl: print error when units were not found in registry
#1566: fleetctl: check also systemd states after starting units
#1604: vendor: bump github.com/vishvananda/netlink to fix build error on Darwin
#1623: fleetd: check that eStream is available before accessing to that
#1633: fleetctl: reenable support for environment variables
#1639: agent: reenable check for nil hash in unit state publisher
#1646: systemd,agent: unload unit only when TriggerStop() runs successfully
#1647: agent,unit: check errors from start/stop/unload-ing unit
#1648: fleetctl: fix error message when fleetctl ssh fails
#1651: agent: make TestUnitStatePublisher wait for each goroutine to finish
#1653: api: reject templates for particular unit types
#1666: fleetctl: support options --no-block, --block-attempts for destroy command
#1667: fleetctl: check systemd active state via client API
#1685: registry,fleetctl: fix bugs regarding shadowed error variables
#1693: registry/rpc: re-enable error handling after NewSystemdUnitManager
#1698: agent,engine: fix bugs in rescheduling for replaced units
#1700: functional: fix error from message change from systemctl status
#1701: fleetctl: take experimentalAPI into account in getClient
#1706: functional: close DBUS connections after running systemd commands
#1713: engine: fix a bug in engine being unreachable
#1717: server: make Monitor shut down the process hard

v0.11.8

2016-09-16T10:00:33Z

New Features

#1630: add support for etcd basic authentication

v0.13.0

2016-10-07T10:13:52Z

Release from tag v0.13.0, commit 1d6e74a ("Merge pull request #1572"), as of 2016-05-30.

New Features

#1509: fleet: add replace unit support
#1560: fleetctl: inform the user about the '-replace' switch in case the units differ
#1524: metrics: added prometheus metrics for fleet
#1572: fleetd: introduce Replaces option in unit files
#1574: fleetctl: support truthy values in boolean unit options
#1577: fleetctl: print out result message explicitly
#1581: fleetctl: use cobra instead of cli
(NOTE: #1581 introduces another bug in supporting environment variables.
That was fixed by #1633, which is not included in v0.13.0.)

Bugs fixed

#1515: fleetd:registry: improve Units() performance when fetching Jobs list from etcd
#1540: ssh: Do not leak ssh sessions on errors
#1561: fleetd: detect the existing machine ID
#1580: registry: use etcd.Config.HeaderTimeoutPerRequest instead of internal timeout

v0.12.0

2016-10-07T10:13:21Z

Release from tag v0.12.0, commit e2a8370 ("Merge pull request #1544"), as of 2016-04-15.

New Features

#1263: Implement follow_only flag
#1265: Add flag to override tokenLimit
#1341: fleetctl: optimize status command
#1406: Add --disable_watches flag
#1416: fleetctl: add --output to journal
#1430: test: added test scripts for the functional tests

Bugs fixed

#1374: fleetd: fix argument handling
#1376: consolidate excessive Hash/Unit requests
#1417: fleetctl: on destroy command ignore units that do not exist
#1475: fleetctl: check for empty input unit strings
#1492: tests: added "fleetctl --tunnel" tests (fixes #1425)
#1496: server: fix panic on graceful shutdown
#1512: tests: added template + metadata functional test (fixes #1446)
#1517: fleetd: keep a list of listeners across reconfiguring fleetd
#1520: fixed broken template and metadata
#1529: ssh: define the list of Key Algorithms of remote hosts before handshake

v0.11.7

2016-04-11T13:20:22Z

Bugs Fixed

prefer key algorithm from known_hosts over the server's (#1526)

v0.11.6

2016-04-05T00:39:00Z

v0.11.6

v0.11.5

2015-09-01T19:07:39Z

Bugs fixed

do not block forever waiting for global unit creation (#1333)

v0.11.4

2015-08-14T22:01:35Z

Bugs fixed

correctly block waiting for unit creation in submit/load/start before proceeding with other actions
block waiting for unit destruction in destroy

v0.11.3

2015-08-14T21:55:44Z

Bugs fixed

reworked race condition fix included in the last release. using consistent reads in fleetd for etcd was deemed too demanding, but watches don't work as expected against etcd 0.4.x. instead we wait for unit creation in fleetctl before proceeding with other actions.

Known bugs

does not correctly wait for unit creation when the unit exists on the local disk but not in the remote registry. this is fixed in release v0.11.4.

v0.11.2

2015-07-23T05:40:52Z

Bugs fixed

fix a race condition when creating units due to etcd reads without consistency