consul/website/source/docs/guides/semaphore.html.md

6.8 KiB
Raw Blame History

layout page_title sidebar_current description
docs Semaphore docs-guides-semaphore This guide demonstrates how to implement a distributed semaphore using the Consul KV store.

Semaphore

This guide demonstrates how to implement a distributed semaphore using the Consul KV store. This is useful when you want to coordinate many services while restricting access to certain resources.

~> If you only need mutual exclusion or leader election, this guide provides a simpler algorithm that can be used instead.

There are a number of ways that a semaphore can be built, so our goal is not to cover all the possible methods. Instead, we will focus on using Consul's support for sessions. Sessions allow us to build a system that can gracefully handle failures.

-> Note: JSON output in this guide has been pretty-printed for easier reading. Actual values returned from the API will not be formatted.

Contending Nodes

Let's imagine we have a set of nodes who are attempting to acquire a slot in the semaphore. All nodes that are participating should agree on three decisions: the prefix in the KV store used to coordinate, a single key to use as a lock, and a limit on the number of slot holders.

For the prefix we will be using for coordination, a good pattern is simply:

service/<service name>

We'll abbreviate this pattern as simply <prefix> for the rest of this guide.

The first step is for each contender to create a session. This is done using the Session HTTP API:

curl  -X PUT -d '{"Name": "db-semaphore"}' \
  http://localhost:8500/v1/session/create

This will return a JSON object contain the session ID:

{
  "ID": "4ca8e74b-6350-7587-addf-a18084928f3c"
}

-> Note: Sessions by default only make use of the gossip failure detector. That is, the session is considered held by a node as long as the default Serf health check has not declared the node unhealthy. Additional checks can be specified at session creation if desired.

Next, we create a lock contender entry. Each contender creates a kv entry that is tied to a session. This is done so that if a contender is holding a slot and fails, its session is detached from the key, which can then be detected by the other contenders.

Create the contender key by doing an acquire on <prefix>/<session> via PUT. This is something like:

curl -X PUT -d <body> http://localhost:8500/v1/kv/<prefix>/<session>?acquire=<session>

body can be used to associate a meaningful value with the contender, such as its nodes name. This body is opaque to Consul but can be useful for human operators.

The <session> value is the ID returned by the call to /v1/session/create.

The call will either return true or false. If true, the contender entry has been created. If false, the contender node was not created; it's likely that this indicates a session invalidation.

The next step is to create a single key to coordinate which holders are currently reserving a slot. A good choice for this lock key is simply <prefix>/.lock. We will refer to this special coordinating key as <lock>.

This is done with:

curl -X PUT -d <body> http://localhost:8500/v1/kv/<lock>?cas=0

Since the lock is being created, a cas index of 0 is used so that the key is only put if it does not exist.

body should contain both the intended slot limit for the semaphore and the session ids of the current holders (initially only of the creator). A simple JSON body like the following works:

{
    "Limit": 2,
    "Holders": [
      "<session>"
    ]
}

The current state of the semaphore is read by doing a GET on the entire <prefix>:

curl http://localhost:8500/v1/kv/<prefix>?recurse

Within the list of the entries, we should find two keys: the <lock> and the contender key /.

[
  {
    "LockIndex": 0,
    "Key": "<lock>",
    "Flags": 0,
    "Value": "eyJMaW1pdCI6IDIsIkhvbGRlcnMiOlsiPHNlc3Npb24+Il19",
    "Session": "",
    "CreateIndex": 898,
    "ModifyIndex": 901
  },
  {
    "LockIndex": 1,
    "Key": "<prefix>/<session>",
    "Flags": 0,
    "Value": null,
    "Session": "<session>",
    "CreateIndex": 897,
    "ModifyIndex": 897
  }
]

Note that the Value we embedded into <lock> is Base64 encoded when returned by the API.

When the <lock> is read and its Value is decoded, we can verify the Limit agrees with the Holders count. This is used to detect a potential conflict. The next step is to determine which of the current slot holders are still alive. As part of the results of the GET, we also have all the contender entries. By scanning those entries, we create a set of all the Session values. Any of the Holders that are not in that set are pruned. In effect, we are creating a set of live contenders based on the list results and doing a set difference with the Holders to detect and prune any potentially failed holders. In this example <session> is present in Holders and is attached to the key <prefix>/<session>, so no pruning is required.

If the number of holders after pruning is less than the limit, a contender attempts acquisition by adding its own session to the Holders list and doing a Check-And-Set update of the <lock>. This performs an optimistic update.

This is done with:

curl -X PUT -d <Updated Lock Body> http://localhost:8500/v1/kv/<lock>?cas=<lock-modify-index>

lock-modify-index is the latest ModifyIndex value known for <lock>, 901 in this example.

If this request succeeds with true, the contender now holds a slot in the semaphore. If this fails with false, then likely there was a race with another contender to acquire the slot.

To re-attempt the acquisition, we watch for changes on <prefix>. This is because a slot may be released, a node may fail, etc. Watching for changes is done via a blocking query against /kv/<prefix>?recurse.

Slot holders must continously watch for changes to <prefix> since their slot can be released by an operator or automatically released due to a false positive in the failure detector. On changes to <prefix> the locks Holders list must be re-checked to ensure the slot is still held. Additionally, if the watch fails to connect the slot should be considered lost.

This semaphore system is purely advisory. Therefore it is up to the client to verify that a slot is held before (and during) execution of some critical operation.

Lastly, if a slot holder ever wishes to release its slot voluntarily, it should be done by doing a Check-And-Set operation against <lock> to remove its session from the Holders object. Once that is done, both its contender key <prefix>/<session> and session should be deleted.