Storage Manager
|
Note |
|---|---|
This page is actively maintained by the Grid'5000 team. If you encounter problems, please report them (see the Support page). Additionally, as it is a wiki page, you are free to make minor corrections yourself if needed. If you would like to suggest a more fundamental change, please contact the Grid'5000 team. | |
The Storage Manager manages access to NFS shares (in particular, users' home directories) on Grid'5000.
This page describes its API, and focuses on accessing user's home directories.
How it works
The storage API works with access object. An access is defined by a list of addresses and a termination condition for a specific location on an NFS Server. Each user's home directory is located on /export/home/username on the NFS server of each site: nfs.site.grid5000.fr
Each time you start a job, an access to your local home directory is automatically created for all the assigned nodes. You may want to use this page if you want to allow additional addresses to access your home (like when you are using VLAN) or if you want to access your home over-site. You will need to use the Grid'5000 API, the entry-point for the home of a user on a specific site is:
https://api.grid5000.fr/stable/sites/site/storage/home/username/access
For Group Storage, the entry-point of a storage on a server server of a site is:
https://api.grid5000.fr/stable/sites/site/storage/server/storage/access
For example, https://api.grid5000.fr/stable/sites/lille/storage/storage1/my-team-storage/access
|
|
Note |
|---|---|
Only the home owner can interact with his or her accesses, so you need to be authenticated. It's done automatically on the frontend. With curl, you can also use the | |
Usage
We assume that we want to manage the home of user jpicard on nancy, so the storage API entry-point will be:
https://api.grid5000.fr/stable/sites/nancy/storage/home/jpicard/access
Create access
Accesses creation is done by sending a POST request to the API with a JSON payload.
We must provide a list of hosts to whom we want to give access to, for instance dahu-1.grenoble.grid5000.fr and dahu-2.grenoble.grid5000.fr. We also need a termination condition, it can be a specific date or a job termination.
|
|
Note |
|---|---|
To give access to a home from a specific VLAN, you can use its domain name, for example for dahu-1 on VLAN 2, you can add | |
Until termination
If we want to use a specific time (for example 2018-12-25 19:38), we send the following request to the Storage API:
curl -X POST 'https://api.grid5000.fr/stable/sites/nancy/storage/home/jpicard/access' -H "Content-Type: application/json" -d '{"hosts": ["dahu-1.grenoble.grid5000.fr", "dahu-2.grenoble.grid5000.fr"], "termination" : {"until":"2018-12-25 19:38"}}'
Job termination
If we want our access to be valid until the end of a Grid'5000 job, for example the job 4548 of grenoble, we do:
curl -X POST 'https://api.grid5000.fr/stable/sites/nancy/storage/home/jpicard/access' -H "Content-Type: application/json" -d '{"hosts": ["dahu-1.grenoble.grid5000.fr", "dahu-2.grenoble.grid5000.fr"], "termination" : {"job":4548, "site":"grenoble"}}'
If we want to give access to all the assigned nodes of a job, we don't need to give any hosts, the API will fill them for us:
curl -X POST 'https://api.grid5000.fr/stable/sites/nancy/storage/home/jpicard/access' -H "Content-Type: application/json" -d '{"termination" : {"job":4548, "site":"grenoble"}}'
|
Warning |
|---|---|
The access is created immediately and is valid until the termination of the job, even if the job has not started yet. | |
The nfs_address field will contain the address used to manually mount the home:
|
node:
|
mount srv-data.nancy.grid5000.fr:/export/home/
jpicard /mnt/ |
|
|
Note |
|---|---|
| |
List access
To list all accesses, you should do:
curl https://api.grid5000.fr/stable/sites/nancy/storage/home/jpicard/access
You get a list of accesses:
{
"G5k-home_jpicard_j_1666466-nancy_1": {
"ipv4": [
"172.16.64.97"
],
"termination": {
"job": 1666466,
"site": "nancy"
},
"nfs_address": "srv-data.nancy.grid5000.fr:/export/home/jpicard"
},
"G5k-home_jpicard_u_1535456240_1": {
"ipv4": [
"172.16.64.16"
],
"termination": {
"until": 1535456240,
},
"nfs_address": "srv-data.nancy.grid5000.fr:/export/home/jpicard"
}
}
You can also ask for a specific access by providing its identifier, for instance:
curl https://api.grid5000.fr/stable/sites/nancy/storage/home/jpicard/access/G5k-home_jpicard_u_1535456240_1
Delete an access
Access deletion is done by sending a DELETE request to the API with the access identifier.
curl -X DELETE https://api.grid5000.fr/stable/sites/nancy/storage/home/jpicard/access/G5k-home_jpicard_u_1535456240_1
API Specification:
The Storage API uses the same status codes as the rest of the Grid'5000 API.
GET /stable/sites/site/storage/home/username/access
Fetch a list of access.
GET /stable/sites/site/storage/home/username/access/access_id
Fetch information of a specific access.
DELETE /stable/sites/site/storage/home/username/access/access_id
Delete a specific access.
DELETE /stable/sites/site/storage/home/username/access/?job=job_id&site=job_site
Delete all access with a job termination job_id of site job_site.
POST /stable/sites/site/storage/home/username/access
Create new access. The payload must be in JSON format, and contains the following properties:
| Parameter | Type | Description & Examples |
|---|---|---|
| ipv4 | Optional | Array of ipv4 allowed to access NFS. Eg:
["172.16.72.1", "172.16.72.25"] |
| hosts | Optional | Array of hosts allowed to access NFS. Eg:
["grisou-1.nancy.grid5000.fr", "grisou-2.nancy.grid5000.fr"] |
| termination | Mandatory | Termination condition can be of two types "Until" or "Job". The form is a structure/hash.
Until termination takes a date as an until parameter. The access will be deleted after the specify date. Eg: {"until" : "2018-12-25 19:38"}
Job terminations takes two parameters, job and site, which are respectively the id and the site of a Grid'5000 job. When the job ends, the export is deleted. eg: {"job": "4548, "site": "nancy"}
If no hosts and no ipv4 are provided, the API will use the assigned nodes from the job. |