Skip to main content

ETCD: POST vs. PUT understanding

ETCD is distributed key value store used as a core component in CoreOS. I've already send a post earlier this week. Here is a page describing how to use ETCD basic commands = ETCD API. Code snippets placed in a page mostly use put, but ETCD allows to use post as well. Most of us understand differences between those two commands in a notion of a REST(ful) service, but how does it work in key value store?

POST

Example over many words.

curl -v http://127.0.0.1:2379/v2/keys/test -XPOST -D value="some value"
curl -v http://127.0.0.1:2379/v2/keys/test -XPOST -D value="some value"

Two same command result into following content:

1
 2
 3
 4
 5
 6
 7
 8
 9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
{
  "action": "get",
  "node": {
    "key": "/test",
    "dir": true,
    "nodes": [
      {
        "key": "/test/194",
        "value": "",
        "modifiedIndex": 194,
        "createdIndex": 194
      },
      {
        "key": "/test/195",
        "value": "",
        "modifiedIndex": 195,
        "createdIndex": 195
      }
    ],
    "modifiedIndex": 194,
    "createdIndex": 194
  }
}

So ETCD adds an index value and put it into resulting key - which is also path to the value. For instance:

curl -v http://127.0.0.1:2379/v2/keys/test/194 -XGET

Allows you to get the specific key. The index is explicitly expressed in the url.

PUT

Put command just add or update given key. Let say I would use following example:

curl -v http://127.0.0.1:2379/v2/keys/test -XPUT -D value="some value"

Resulting content on test key is expected.

1
2
3
4
5
6
7
8
9
{
  "action": "get",
  "node": {
    "key": "/test",
    "value": "",
    "modifiedIndex": 198,
    "createdIndex": 198
  }
}

How to Model Add and Update Method?

My current task is to model and implement repository using ETCD under the hood. Usual repository contains CRUD methods for particular set of entities. Reasonable approach is to separate add from update to do not replace existing object, e.g. when using optimistic locking.

I don't want to see revision - index - numbers within keys so post command is not useful here. ETCD brings prevExist parameter for this use cases.

I want to perform add method which expect that there is no content on given key. I'll use following statement:

curl -v http://127.0.0.1:2379/v2/keys/test?prevExist=false -XPUT -D value="some value"

When you did not delete the key, as I did not, you can get following error:

1
2
3
4
5
6
{
  "errorCode": 105,
  "message": "Key already exists",
  "cause": "/test",
  "index": 198
}

On the other hand, use false to express update existing entity.

curl -v http://127.0.0.1:2379/v2/keys/test?prevExist=true -XPUT -D value="some value"

This command results into positive response.

< HTTP/1.1 200 OK

The repository uses put for both add and update methods but value for prevExist is the difference.
1 comment

Popular posts from this blog

Http and TCP Load Balancing with Kubernetes

Kubernetes allows to manage large clusters which are composed of docker containers. And where is large computation power there is large amount of data throughput. However, there is still a need to spread the data throughput so it can reach and utilize particular docker container(s). This is called load balancingKubernetes supports two basic forms of load balancing. Internal among docker containers themselves and external for clients which call you application from outside environment - just users.


Internal Load Balancing with Kubernetes Usual approach during the modeling of an application in kubernetes is to provide domain models for pods, replications controllers and services. Look at the great documentation if you are not aware of all these principles.

For simplification, pod is a set of docker containers which are always located on the same node. Replication controller allows and guarantees scalability. The last but not least is the service which hides all instances of pods - cre…

Validating nginx config file using docker approach

I try to setup nginx as a load balancer. The configuration is just a file with a lot of constrains so I need a validation. There is no online validation service, as e.g. CoreOS has, and I don't want to install nginx on my laptop as I work on a distributed app.

Docker is right approach for me. Let say I have following config:


In short, I'm going to pass nginx config to running nginx instance and look to the logs.

Put you nginx.config to the temp and start the docker image:

sudo docker run --name nginx -v /tmp/nginx.config:/etc/nginx/nginx.conf:ro -d nginx It uses volume mapping so the command just starts a new docker container and mounts a local /tmp/nginx.config to the given in-container path. You can obviously change the volume path to your personal path. Is it working or not? Look at logs.

sudo docker logs nginx If there is no entry, your file is fine. In the case of an error, you can see something like this:

2016/01/08 11:37:31 [emerg] 1#1: unexpected "}" in /etc/…