README.md 21 KB
Newer Older
1
# Operating an OASIS
2

Aviral Vaid's avatar
Aviral Vaid committed
3
4
5
Originally, NOMAD Central Repository is a service run at Max-Planck's computing facility in Garching, Germany.
However, the NOMAD software is Open-Source, and everybody can run it. Any service that
uses NOMAD software independently is called a *NOMAD OASIS*.
6

Aviral Vaid's avatar
Aviral Vaid committed
7
While several use cases require different setups, this documentation
8
describes the simplest setup of a NOMAD OASIS. It would allow a group to use NOMAD for
Aviral Vaid's avatar
Aviral Vaid committed
9
local research data management, while using NOMAD's central user-management and its users.
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
There are several environment in which you can run a NOMAD OASIS: base-metal linux,
docker with docker-compose, and in a kubernetes cluster. We recommend using docker/docker-compose.

## Before you start

We need to register some information about your OASIS in the central user management:
- The hostname for the machine you run NOMAD on. This is important for redirects between
your OASIS and the central NOMAD user-management and to allow your users to upload files (via GUI or API).
Your machine needs to be accessible under this hostname from the public internet. The host
name needs to be registered with the central NOMAD in order to configure the central user-
management correctly.
- Your NOMAD account that should act as an admin account for your OASIS. This account must be declared
to the central NOMAD as an OASIS admin in order to give it the necessary rights in the central user-
management.
- You will need to know your NOMAD user-id. This information has to be provided by us.

Please [write us](mailto:support@nomad-lab.eu) to register your NOMAD account as an OASIS
admin and to register your hostname. Please replace the indicated configuration items with
the right information.


In principle, you can also run your own user management. This is not yet documented.
32

33
34
35
## docker and docker compose

### Pre-requisites
36
37
38
39
40
41
42
43
44
45
46
47

NOMAD software is distributed as a set of docker containers. Further, other services
that can be run with docker are required. Further, we use docker-compose to setup
all necessary container in the simples possible manner.

You will need a single computer, with **docker** and **docker-compose** installed.

The following will run all necessary services with docker. These comprise: a **mongodb**
database, an **elasticsearch**, a **rabbitmq** distributed task queue, the NOMAD **app**,
NOMAD **worker**, and NOMAD **gui**. Refer to this [introduction](/app/docs/introduction.html#architecture)
to learn what each service does and why it is necessary.

48
### Configuration overview
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
66
67
68
69
70
71
72
73
74
75
76
77
78
79
80
81
82
83
84
85
86
87
88
89
90
91
92
93

All docker container are configured via docker-compose an the respective `docker-compose.yaml` file.
Further, we will need to mount some configuration files to configure the NOMAD services within
their respective containers.

There are three files to configure:
- `docker-compose.yaml`
- `nomad.yaml`
- `nginx.conf`

In this example, we have all files in the same directory (the directory we also work from).
You can download examples files from
[here](https://gitlab.mpcdf.mpg.de/nomad-lab/nomad-FAIR/tree/master/ops/docker-compose/nomad-oasis/).

### Docker compose

The most basic `docker-compose.yaml` to run an OASIS looks like this:

```yaml
version: '3.4'

x-common-variables: &nomad_backend_env
    NOMAD_RABBITMQ_HOST: rabbitmq
    NOMAD_ELASTIC_HOST: elastic
    NOMAD_MONGO_HOST: mongo

services:
    # broker for celery
    rabbitmq:
        restart: always
        image: rabbitmq:3.7.17
        container_name: nomad_oasis_rabbitmq
        environment:
            - RABBITMQ_ERLANG_COOKIE=SWQOKODSQALRPCLNMEQG
            - RABBITMQ_DEFAULT_USER=rabbitmq
            - RABBITMQ_DEFAULT_PASS=rabbitmq
            - RABBITMQ_DEFAULT_VHOST=/
        volumes:
            - nomad_oasis_rabbitmq:/var/lib/rabbitmq

    # the search engine
    elastic:
        restart: always
        image: docker.elastic.co/elasticsearch/elasticsearch:6.3.2
        container_name: nomad_oasis_elastic
94
95
        environment:
            - discovery.type=single-node
96
97
98
99
100
101
102
103
104
105
106
107
108
109
110
111
112
113
        volumes:
            - nomad_oasis_elastic:/usr/share/elasticsearch/data

    # the user data db
    mongo:
        restart: always
        image: mongo:4
        container_name: nomad_oasis_mongo
        environment:
            - MONGO_DATA_DIR=/data/db
            - MONGO_LOG_DIR=/dev/null
        volumes:
            - nomad_oasis_mongo:/data/db
        command: mongod --logpath=/dev/null # --quiet

    # nomad worker (processing)
    worker:
        restart: always
114
        image: gitlab-registry.mpcdf.mpg.de/nomad-lab/nomad-fair:stable
115
116
117
118
119
120
121
122
123
124
        container_name: nomad_oasis_worker
        environment:
            <<: *nomad_backend_env
            NOMAD_SERVICE: nomad_oasis_worker
        links:
            - rabbitmq
            - elastic
            - mongo
        volumes:
            - ./nomad.yaml:/app/nomad.yaml
125
            - nomad_oasis_files:/app/.volumes/fs
126
127
        command: python -m celery worker -l info -A nomad.processing -Q celery,calcs,uploads

128
    # nomad app (api + gui)
129
130
    app:
        restart: always
131
        image: gitlab-registry.mpcdf.mpg.de/nomad-lab/nomad-fair:stable
132
133
134
135
136
137
138
139
140
141
        container_name: nomad_oasis_app
        environment:
            <<: *nomad_backend_env
            NOMAD_SERVICE: nomad_oasis_app
        links:
            - rabbitmq
            - elastic
            - mongo
        volumes:
            - ./nomad.yaml:/app/nomad.yaml
142
            - nomad_oasis_files:/app/.volumes/fs
143
        command: ./run.sh
144

145
    # nomad gui (a reverse proxy for nomad)
146
147
    gui:
        restart: always
148
        image: nginx:1.13.9-alpine
149
150
151
152
153
154
155
156
157
158
159
160
161
162
163
164
165
166
167
168
        container_name: nomad_oasis_gui
        command: nginx -g 'daemon off;'
        volumes:
            - ./nginx.conf:/etc/nginx/conf.d/default.conf
        links:
            - app
        ports:
            - 80:80

volumes:
    nomad_oasis_mongo:
    nomad_oasis_elastic:
    nomad_oasis_rabbitmq:
    nomad_oasis_files:
```

There are no mandatory changes necessary.

A few things to notice:
- All services use docker volumes for storage. This could be changed to host mounts.
169
- It mounts three configuration files that need to be provided (see below): `nomad.yaml`, `nginx.conf`, `env.js`.
170
171
172
173
174
175
176
177
178
179
180
181
182
183
184
185
- The only exposed port is `80`. This could be changed to a desired port if necessary.
- The NOMAD images are pulled from our gitlab in Garching, the other services use images from a public registry (*dockerhub*).
- All container will be named `nomad_oasis_*`. These names can be used to later reference the container with the `docker` cmd.
- The NOMAD images we use are tagged `stable`. This could be replaced with concrete version tags.
- The services are setup to restart `always`, you might want to change this to `no` while debugging errors to prevent
indefinite restarts.

### nomad.yaml

NOMAD app and worker read a `nomad.yaml` for configuration.

```yaml
client:
  url: 'http://<your-host>/nomad-oasis/api'

services:
186
  api_prefix: '/nomad-oasis'
187
188
189
190
191
192
193
  admin_user_id: '<your admin user id>'

keycloak:
  realm_name: fairdi_nomad_prod
  username: '<your admin username>'
  password: '<your admin user password>'
  oasis: true
194
195
196
197
198
199

mongo:
    db_name: nomad_v0_8

elastic:
    index_name: nomad_v0_8
200
201
202
203
204
205
206
207
208
209
210
```

You need to change:
- Replace `your-host` and admin credentials respectively.
- `api_base_path` defines the path under with the app is run. It needs to be changed, if you use a different base path.

A few things to notice:
- Be secretive about your admin credentials; make sure this file is not publicly readable.

### nginx.conf

211
The GUI container serves as a proxy that forwards request to the app container. The
212
213
proxy is an nginx server and needs a configuration similar to this:

214
```none
215
216
server {
    listen        80;
217
    server_name   www.example.com;
218
219
    proxy_set_header Host $host;

220
    location / {
221
222
223
224
        proxy_pass http://app:8000;
    }

    location ~ /nomad-oasis\/?(gui)?$ {
225
        rewrite ^ /nomad-oasis/gui/ permanent;
226
227
228
229
230
231
232
    }

    location /nomad-oasis/gui/ {
        proxy_intercept_errors on;
        error_page 404 = @redirect_to_index;
        proxy_pass http://app:8000;
    }
233

234
    location @redirect_to_index {
235
        rewrite ^ /nomad-oasis/gui/index.html break;
236
237
238
        proxy_pass http://app:8000;
    }

239
    location ~ \/gui\/(service-worker\.js|meta\.json)$ {
240
241
242
243
244
        add_header Last-Modified $date_gmt;
        add_header Cache-Control 'no-store, no-cache, must-revalidate, proxy-revalidate, max-age=0';
        if_modified_since off;
        expires off;
        etag off;
245
        proxy_pass http://app:8000;
246
247
    }

248
    location ~ \/api\/uploads\/?$ {
249
250
251
252
253
        client_max_body_size 35g;
        proxy_request_buffering off;
        proxy_pass http://app:8000;
    }

254
255
256
257
258
259
    location ~ \/api\/(raw|archive) {
        proxy_buffering off;
        proxy_pass http://app:8000;
    }

    location ~ \/api\/mirror {
260
        proxy_buffering off;
261
        proxy_read_timeout 600;
262
263
264
265
266
267
268
269
270
271
272
        proxy_pass http://app:8000;
    }

    location ~ \/encyclopedia\/ {
        proxy_intercept_errors on;
        error_page 404 = @redirect_to_encyclopedia_index;
        proxy_pass http://app:8000;
    }

    location @redirect_to_encyclopedia_index {
        rewrite ^ /nomad-oasis/encyclopedia/index.html break;
273
274
275
276
277
278
279
280
281
282
283
284
285
286
        proxy_pass http://app:8000;
    }
}
```

You need to change:
- Replace `<your-host>`

A few things to notice:
- It configures the base path (`nomad-oasis`) at multiple places. It needs to be changed, if you use a different base path.
- You can use the server to server additional content if you like.
- `client_max_body_size` sets a limit to the possible upload size.
- If you operate the GUI container behind another proxy, keep in mind that your proxy should not buffer requests/responses to allow streaming of large requests/responses for `../api/uploads` and `../api/raw`.

287
288
### gunicorn

289
290
291
Gunicorn is the WSGI-server that runs the nomad app. Consult the
[gunicorn documentation](https://docs.gunicorn.org/en/stable/configure.html) for
configuration options.
292

293
### Starting and stopping NOMAD
294
295

If you prepared the above files, simply use the usual `docker-compose` commands to start everything.
296
297

To make sure you have the latest docker images for everything run this first:
298
```sh
299
300
301
docker-compose pull
```

302
In the beginning and for debugging problems, it is recommended to start services separately:
303
```sh
304
305
306
307
308
docker-compose up -d mongodb elastic rabbitmq
docker-compose up app worker gui
```

The `-d` option runs container in the background as *daemons*. Later you can run all at once:
309
```sh
310
311
312
313
docker-compose up -d
```

You can also use docker to stop and remove faulty containers that run as *daemons*:
314
```sh
315
316
317
318
319
docker stop nomad_oasis_app
docker rm nomad_oasis_app
```

If everything works, the gui should be available under:
320
```none
321
322
323
324
325
326
327
http://<your host>/nomad-oasis/gui/
```

If you run into troubles, use the dev-tools of you browser to check the javascript logs
or monitor the network traffic for HTTP 500/400/404/401 responses.

To see if at least the api works, check
328
```none
329
330
331
332
333
334
335
http://<your host>/nomad-oasis/alive
http://<your host>/nomad-oasis/api/info
```

To see logs or 'go into' a running container, you can access the individual containers
with their names and the usual docker commands:

336
```sh
337
338
339
docker logs nomad_oasis_app
```

340
```sh
341
342
343
344
345
346
docker exec -ti nomad_oasis_app /bin/bash
```

If you want to report problems with your OASIS. Please provide the logs for
- nomad_oasis_app
- nomad_oasis_worker
347
348
- nomad_oasis_gui

349
350
351
352
353
## base linux (without docker)

### Pre-requisites

We will run NOMAD from the *nomad-lab* Python package. This package contains all the necessary
354
code for running the processing, api, and gui. But, it is missing the necessary databases. You might be
355
356
357
358
359
360
361
362
363
364
365
366
367
able to run NOMAD in user space.

You will need:

- preferably a linux computer, which Python 3.7, preferable a Python virtual environment
- elasticsearch 6.x, running without users and authentication, preferable on the default settings
- mongodb 4.x, running without users and authentication, preferable on the default settings
- rabbitmq 3.x, running without users and authentication, preferable on the default settings
- nginx
- an empty directory to work in

### Install the NOMAD Python package

368
You should install everything in a virtual environment. For example like this:
369
370
371
372
373
```sh
virtualenv -p `which python3` nomadpyenv
source nomadpyenv/bin/activate
```

374
You can simply install the Python package from pypi:
375
376
377
378
```sh
pip install nomad-lab[all]
```

379
380
If you need the latest version, you can also download the latest package from our
"beta" installation.
381
382
383
384
385
386
387
```sh
curl "https://nomad-lab.eu/prod/rae/beta/dist/nomad-lab.tar.gz" -o nomad-lab.tar.gz
pip install nomad-lab.tar.gz[all]
```

### Configure NOMAD - nomad.yaml

388
389
The `nomad.yaml` is our central config file. You should write a `nomad.yaml` like this:

390
391
392
393
394
395
396
397
398
399
400
401
402
403
404
405
406
407
408
409
410
411
412
413
414
415
416
417
418
419
420
421
422
423
424
425
426
427
428
429
430
431
432
433
434
435
436
437
438
439
440
441
442
443
444
445
446
447
448
449
450
451
452
453
454
455
456
457
458
459
460
```yaml
client:
  url: 'http://<your-host>/nomad-oasis/api'

services:
  api_base_path: '/nomad-oasis'
  admin_user_id: '<your admin user id>'

keycloak:
  realm_name: fairdi_nomad_prod
  username: '<your admin username>'
  password: '<your admin user password>'
  oasis: true

mongo:
    db_name: nomad_v0_8

elastic:
    index_name: nomad_v0_8
```

You need to change:
- Replace `your-host` and admin credentials respectively.
- `api_base_path` defines the path under with the app is run. It needs to be changed, if you use a different base path.

A few things to notice:
- Be secretive about your admin credentials; make sure this file is not publicly readable.

### Configure NOMAD - nginx

You can generate a suitable `nginx.conf` with the `nomad` command line command that
comes with the *nomad-lab* Python package. If you server other content but NOMAD with
your nginx, you need to incorporate the config accordingly.

If you have a standard installation of nginx, this might work. Adapt to your set-up:
```sh
nomad admin ops nginx-conf > /etc/nginx/conf.d/default.conf
nginx -t
nginx -s reload
```

If you want to run nginx in docker, this might work. Adapt to your set-up:
```sh
nomad admin ops nginx-conf --host host.docker.internal > nginx.conf
docker run --rm -v `pwd`/nginx.conf:/etc/nginx/conf.d/default.conf -p 80:80 nginx:stable nginx -g 'daemon off;' &
```

### Running NOMAD

Before you start, we need to transfer your `nomad.yaml` config values to the GUI's
javascript. You need to repeat this, if you change your `nomad.yaml`. You can do this by running:
```
nomad admin ops gui-config
```

To run NOMAD, you must run two services. One is the NOMAD app, it serves the API and GUI:
```
gunicorn "${params[@]}" -b 0.0.0.0:8000 nomad.app:app
```

And the NOMAD worker, that runs the NOMAD processing.
```
celery worker -l info -A nomad.processing -Q celery,calcs,uploads
```

This should give you a working OASIS at `http://<your-host>/<your-path-prefix>`.

## kubernetes

*This is not yet documented.*

461
462
## Migrating from an older version (0.7.x to 0.8.x)

463
464
*This documentation is outdated and will be removed in future releases.*

465
466
467
468
469
Between versions 0.7.x and 0.8.x we needed to change how archive and metadata data is stored
internally in files and databases. This means you cannot simply start a new version of
NOMAD on top of the old data. But there is a strategy to adapt the data.

First, shutdown the OASIS and remove all old container.
470
```sh
471
472
473
474
docker-compose stop
docker-compose rm -f
```

475
Update you config files (`docker-compose.yaml`, `nomad.yaml`, `nginx.conf`) according
476
477
478
479
480
to the latest documentation (see above). Especially make sure to select a new name for
databases and search index in your `nomad.yaml` (e.g. `nomad_v0_8`). This will
allow us to create new data while retaining the old, i.e. to copy the old data over.

Make sure you get the latest images and start the OASIS with the new version of NOMAD:
481
```sh
482
483
484
485
486
487
488
489
490
491
492
docker-compose pull
docker-compose up -d
```

If you go to the GUI of your OASIS, it should now show the new version and appear empty,
because we are using a different database and search index now.

To migrate the data, we created a command that you can run within your OASIS' NOMAD
application container. This command takes the old database name as argument, it will copy
all data over and then reprocess the data to create data in the new archive format and
populate the search index. The default database name in version 0.7.x installations was `nomad_fairdi`.
493
Be patient.
494

495
```sh
496
docker exec -ti nomad_oasis_app bash -c 'nomad admin migrate --mongo-db nomad_fairdi'
497
498
499
500
501
```

Now all your data should appear in your OASIS again. If you like, you can remove the
old index and database:

502
```sh
503
docker exec nomad_oasis_elastic bash -c 'curl -X DELETE http://elastic:9200/nomad_fairdi'
504
505
506
docker exec nomad_oasis_mongo bash -c 'mongo nomad_fairdi --eval "printjson(db.dropDatabase())"'
```

507
508
509
510
511
512
513
514
515
516
517
518
519
520
521
522
523
524
525
526
527
528
## Restricting access to your Oasis

An Oasis works exactly the same way the official NOMAD works. It is open and everybody
can access published data. Everybody with an account can upload data. This might not be
what you want.

Currently there are two ways to restrict access to your Oasis. First, you do not
expose the Oasis to the public internet, e.g. you only make it available on an intra-net or
through a VPN.

Second, we offer a simple white-list mechanism. As the Oasis administrator your provide a
list of accounts as part of your Oasis configuration. To use the Oasis, all users have to
be logged in and be on your white list of allowed users. To enable white-listing, you
can provide a list of NOMAD account email addresses in your `nomad.yaml` like this:

```
oasis:
    allowed_users:
        - user1@gmail.com
        - user2@gmail.com
```

529
530
531
532
533
534
535
536
537
538
539
540
541
542
543
544
545
546
547
548
549
550
551
552
553
554
555
556
557
558
559
560
## NOMAD Oasis FAQ

### Why use an Oasis?
There are three reasons: You want to manage data in private, you want local availability of public NOMAD data without network restrictions, or you want to manage large amounts of low quality and preliminary data that is not intended for publication.

### How to organize data in NOMAD Oasis?

#### How can I categorize or label my data?
Current, NOMAD supports the following mechanism to organize data:
data always belongs to one upload, one uploading user, and is assigned an upload datetime; uploads can have a custom name
data can be assigned to multiple independent datasets
data can hold a proprietary id called “external_id”
data can be assigned multiple authors in addition to the uploading user
The next NOMAD release (0.8.x) will contain more features to filter data based on uploader and upload time. It will also include a revised search bar that makes it easier to filter for external_id or upload_name.

####  Is there some rights-based visibility?
No. Currently, NOMAD only supports uploader controlled visibility. The uploader decides when to make an upload public (with or without embargo). The embargo can be used to limit the visibility of an upload to users that the uploader want to share his upload with.

### How to share data with the central NOMAD?

Keep in mind, it is not entirely clear, how we will do this.

#### How to designate Oasis data for publishing to NOMAD?
Currently, you should use one of the organizational mechanism to designate data for being published to NOMAD. we, you can use a dedicated dataset for publishable data.

#### How to upload?

Will will probably provide functionality in the API of the central NOMAD to upload data from an Oasis to the central NOMAD. We will provide the necessary scripts and detailed instructions. Most likely the data that is uploaded to the central NOMAD can be selected via a search query. Therefore, using a dedicated dataset, would be an easy to select criteria.

### How to maintain an Oasis installation?

#### How to install a NOMAD Oasis?
561
Follow our guide: https://nomad-lab.eu/prod/rae/docs/ops.html#operating-a-nomad-oasis
562
563
564
565
566
567
568
569
570
571
572
573
574
575
576

#### How do version numbers work?
There are still a lot of thing in NOMAD that are subject to change. Currently, changes in the minor version number (0.x.0) designate major changes that require data migration. Changes in the patch version number (0.7.x) just contain minor changes and fixes and do not require data migration. Once we reach 1.0.0, NOMAD will use the regular semantic versioning conventions.

#### How to upgrade a NOMAD Oasis?
When we release a new version of the NOMAD software, it will be available as a new Docker image with an increased version number. You simply change the version number in your docker-compose.yaml and restart.

#### What about major releases?
Going from NOMAD 0.7.x to 0.8.x will require data migration. This means the layout of the data has changed and the new version cannot be used on top of the old data. This requires a separate installation of the new version and mirroring the data from the old version via NOMAD’s API. Detailed instructions will be made available with the new version.

#### How to move data between installations?
We the release of 0.8.x, we will clarify and how to move data between installations. (See last question)

#### How to backup my Oasis?
To backup your Oasis at least the file data and mongodb data needs to be backed up. You determined the path to your file data (your uploads) during the installation. This directory can be backed up like any other file backup (e.g. rsync). To backup the mongodb, please refer to the official mongodb documentation: https://docs.mongodb.com/manual/core/backups/. We suggest a simple mongodump export that is backed up alongside your files. The elasticsearch contents can be reproduced with the information in files and mongodb.