rebootlockd

rebootlockd is a reboot manager for the CoreOS update engine which uses etcd to ensure that only a subset of a cluster of machines is rebooting at any given time.

Usage

Listing the Holders

$ rebootlockctl status
Available: 0
Max: 1

MACHINE ID
69d27b356a94476da859461d3a3bc6fd

Unlock Holders

In some cases a machine may go away permanently or semi-permanently while holding a reboot lock. A system administrator can clear this lock using the unlock command.

$ rebootlockctl unlock 69d27b356a94476da859461d3a3bc6fd

Maximum Sempahore

By default the reboot lock only allows a single holder. However, a user may want more than a single machine to be upgrading at a time. This can be done by increasing the semaphore count.

$ rebootlockctl set-max 4
Old: 1
New: 4

Keyspace

Semaphore

Key: coreos.com/updateengine/rebootlock/semaphore

The semaphore is a json document that describes a simple semaphore that clients swap to take the lock. When it is first created it will be initialized like so:

{
	"semaphore": 1,
	"max": 1,
	"holders": []
}

To take the lock a client the document will be swaped with this:

{
	"semaphore": 0,
	"max": 1,
	"holders": [
		"69d27b356a94476da859461d3a3bc6fd"
	]
}

Holders Directory

Key: coreos.com/updateengine/rebootlock/holders/<machineID>

When a rebootlock client takes the lock it should write information about itself to the holders directory. This should be a JSON document with the following information:

{
	"machineID": "69d27b356a94476da859461d3a3bc6fd",
	"startTime": 1397496396
}

This information is used to show an admin who is holding the semaphore and to resolve problems. The holder must delete themselves with a swap operation before incrementing the semaphore. This lets an administrator safely clear a reboot lock for a client that experienced a power outage or other failure before being able to release the lock.

jonboulle/focaccia