generated from interTwin-eu/repository-template
-
Notifications
You must be signed in to change notification settings - Fork 0
/
Copy pathjupyter_template.py
800 lines (669 loc) · 32.8 KB
/
jupyter_template.py
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
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
94
95
96
97
98
99
100
101
102
103
104
105
106
107
108
109
110
111
112
113
114
115
116
117
118
119
120
121
122
123
124
125
126
127
128
129
130
131
132
133
134
135
136
137
138
139
140
141
142
143
144
145
146
147
148
149
150
151
152
153
154
155
156
157
158
159
160
161
162
163
164
165
166
167
168
169
170
171
172
173
174
175
176
177
178
179
180
181
182
183
184
185
186
187
188
189
190
191
192
193
194
195
196
197
198
199
200
201
202
203
204
205
206
207
208
209
210
211
212
213
214
215
216
217
218
219
220
221
222
223
224
225
226
227
228
229
230
231
232
233
234
235
236
237
238
239
240
241
242
243
244
245
246
247
248
249
250
251
252
253
254
255
256
257
258
259
260
261
262
263
264
265
266
267
268
269
270
271
272
273
274
275
276
277
278
279
280
281
282
283
284
285
286
287
288
289
290
291
292
293
294
295
296
297
298
299
300
301
302
303
304
305
306
307
308
309
310
311
312
313
314
315
316
317
318
319
320
321
322
323
324
325
326
327
328
329
330
331
332
333
334
335
336
337
338
339
340
341
342
343
344
345
346
347
348
349
350
351
352
353
354
355
356
357
358
359
360
361
362
363
364
365
366
367
368
369
370
371
372
373
374
375
376
377
378
379
380
381
382
383
384
385
386
387
388
389
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
461
462
463
464
465
466
467
468
469
470
471
472
473
474
475
476
477
478
479
480
481
482
483
484
485
486
487
488
489
490
491
492
493
494
495
496
497
498
499
500
501
502
503
504
505
506
507
508
509
510
511
512
513
514
515
516
517
518
519
520
521
522
523
524
525
526
527
528
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
561
562
563
564
565
566
567
568
569
570
571
572
573
574
575
576
577
578
579
580
581
582
583
584
585
586
587
588
589
590
591
592
593
594
595
596
597
598
599
600
601
602
603
604
605
606
607
608
609
610
611
612
613
614
615
616
617
618
619
620
621
622
623
624
625
626
627
628
629
630
631
632
633
634
635
636
637
638
639
640
641
642
643
644
645
646
647
648
649
650
651
652
653
654
655
656
657
658
659
660
661
662
663
664
665
666
667
668
669
670
671
672
673
674
675
676
677
678
679
680
681
682
683
684
685
686
687
688
689
690
691
692
693
694
695
696
697
698
699
700
701
702
703
704
705
706
707
708
709
710
711
712
713
714
715
716
717
718
719
720
721
722
723
724
725
726
727
728
729
730
731
732
733
734
735
736
737
738
739
740
741
742
743
744
745
746
747
748
749
750
751
752
753
754
755
756
757
758
759
760
761
762
763
764
765
766
767
768
769
770
771
772
773
774
775
776
777
778
779
780
781
782
783
784
785
786
787
788
789
790
791
792
793
794
795
796
797
798
799
800
#!/usr/bin/env python
# -*- coding: utf-8 -*-
import os
import socket
import json
from oauthenticator.oauth2 import OAuthenticator
from oauthenticator.generic import GenericOAuthenticator
from tornado import gen
# Fix for latest jhub server version
from tornado.httpclient import AsyncHTTPClient
import kubespawner
import subprocess
import warnings
import pprint
import jwt
import os
import asyncio
import kubernetes_asyncio as k8s
import nest_asyncio
callback_url = __CALLBACK_URL__
iam_server = __IAM_SERVER__
client_id = __CLIENT_ID__
client_secret = __CLIENT_SECRET__
cookie_secret_str = __COOKIE_SECRET__
jhub_host = __JHUB_HOST__
jhub_ip = __JHUB_IP__
jhub_port = __JHUB_PORT__
jhub_api_url = __JHUB_API_URL__
cookie_secret_bytes = cookie_secret_str.encode('utf-8')
os.environ["OAUTH_CALLBACK"] = callback_url
os.environ['JUPYTERHUB_OAUTH_ACCESS_SCOPES'] = 'none'
cache_file = './iam_secret'
cache_results = {
"client_id": client_id,
"client_secret": client_secret
}
with open(cache_file, "w") as w:
json.dump(cache_results, w)
client_id = cache_results["client_id"]
client_secret = cache_results["client_secret"]
class EnvAuthenticator(GenericOAuthenticator):
@gen.coroutine
def pre_spawn_start(self, user, spawner):
auth_state = yield user.get_auth_state()
pprint.pprint(auth_state)
if not auth_state:
# user has no auth state
return
# define some environment variables from auth_state
self.log.info(auth_state)
spawner.environment['IAM_SERVER'] = iam_server
spawner.environment['IAM_CLIENT_ID'] = client_id
spawner.environment['IAM_CLIENT_SECRET'] = client_secret
spawner.environment['ACCESS_TOKEN'] = auth_state['access_token']
spawner.environment['REFRESH_TOKEN'] = auth_state['refresh_token']
spawner.environment['USERNAME'] = auth_state['oauth_user']['preferred_username']
spawner.environment['JUPYTERHUB_ACTIVITY_INTERVAL'] = "15"
spawner.environment['SSH_NAMESPACE'] = os.environ.get("SSH_NAMESPACE")
amIAllowed = False
groups = jwt.decode(auth_state["access_token"], options={"verify_signature": False, "verify_aud": False})["groups"]
#groups = [s[1:] for s in groups]
if os.environ.get("OAUTH_GROUPS"):
spawner.environment['GROUPS'] = " ".join(groups)
allowed_groups = os.environ["OAUTH_GROUPS"].split(" ")
amIAllowed = any(gr in groups for gr in allowed_groups)
else:
amIAllowed = True
if not amIAllowed:
self.log.error(
"OAuth user contains not in group the allowed groups %s" % allowed_groups
)
raise Exception("OAuth user not in the allowed groups %s" % allowed_groups)
async def authenticate(self, handler, data=None):
code = handler.get_argument("code")
#http_client = self.http_client()
# Fix for latest jhub server version
http_client = AsyncHTTPClient()
params = dict(
redirect_uri=self.get_callback_url(handler),
code=code,
grant_type='authorization_code',
)
params.update(self.extra_params)
headers = self._get_headers()
#token_resp_json = await self._get_token(http_client, headers, params)
# Fix for latest jhub server version
token_resp_json = await self._get_token(headers, params)
#user_data_resp_json = await self._get_user_data(http_client, token_resp_json)
# Fix for latest jhub server version
user_data_resp_json = await self._get_user_data(token_resp_json)
if callable(self.username_key):
name = self.username_key(user_data_resp_json)
else:
name = user_data_resp_json.get(self.username_key)
if not name:
self.log.error(
"OAuth user contains no key %s: %s", self.username_key, user_data_resp_json
)
return
auth_state = self._create_auth_state(token_resp_json, user_data_resp_json)
print("user info ", jwt.decode(auth_state["access_token"], options={"verify_signature": False, "verify_aud": False} ) )
groups = jwt.decode(auth_state["access_token"], options={"verify_signature": False, "verify_aud": False})["groups"]
is_admin = False
if os.environ.get("ADMIN_OAUTH_GROUPS") in groups:
self.log.info("%s : %s is in %s" , (name, os.environ.get("ADMIN_OAUTH_GROUPS"), groups))
is_admin = True
else:
self.log.info(" %s is not in admin group ", name)
return {
'name': name,
'admin': is_admin,
'auth_state': auth_state #self._create_auth_state(token_resp_json, user_data_resp_json)
}
c.JupyterHub.tornado_settings = {'max_body_size': 1048576000, 'max_buffer_size': 1048576000}
c.JupyterHub.log_level = 30
c.JupyterHub.hub_connect_ip = jhub_ip
c.JupyterHub.api_url = jhub_api_url
c.JupyterHub.cookie_secret = cookie_secret_bytes
c.JupyterHub.authenticator_class = EnvAuthenticator
c.Spawner.default_url = '/lab'
c.GenericOAuthenticator.oauth_callback_url = callback_url
c.GenericOAuthenticator.client_id = client_id
c.GenericOAuthenticator.client_secret = client_secret
c.GenericOAuthenticator.authorize_url = iam_server.strip('/') + '/authorize'
c.GenericOAuthenticator.token_url = iam_server.strip('/') + '/token'
c.GenericOAuthenticator.userdata_url = iam_server.strip('/') + '/userinfo'
c.GenericOAuthenticator.scope = ['openid', 'profile', 'email', 'address', 'offline_access', 'groups']
c.GenericOAuthenticator.username_key = "preferred_username"
c.GenericOAuthenticator.enable_auth_state = True
c.Authenticator.allow_all = True
class CustomSpawner(kubespawner.KubeSpawner):
def __init__(self, *args, **kwargs):
super().__init__(*args, **kwargs)
self.map_node_gpu = {}
self.map_node_fpga = {}
self.gpus_status = {}
self.fpga_status = {}
self.notebook_dir = ""
self.gpus_model_known = [
{
"model": "T4",
"image_url": "https://s.alicdn.com/@sc04/kf/Hb227c34878ab414388e8206abd03ef3cd.jpg_720x720q50.jpg"
},
{
"model": "A100",
"image_url": "https://www.workstationshop.it/pimages/Nvidia-A100-40GB-HBM2-extra-big-16862.png"
},
{
"model": "A200",
"image_url": "https://www.workstationshop.it/pimages/Nvidia-A100-40GB-HBM2-extra-big-16862.png"
}
]
self.fpgas_model_known = [
{
"model": "U55C",
"image_url": "https://www.colfaxdirect.com/store/pc/catalog/u55c.png"
}
]
def get_args(self):
# Get the default arguments
args = super().get_args()
if self.image == "biancoj/jlab-ai":
args.extend([
"/opt/conda/bin/python3",
"/usr/local/bin/jupyterhub-singleuser"
])
#self.notebook_dir = "/home/jovyan"
else: #self.image == "ghcr.io/dodas-ts/htc-dask-wn:v1.0.6-ml-infn-ssh-v5":
args.extend([
"/opt/ssh/jupyterhub-singleuser"
])
#self.notebook_dir = "/jupyter-workspace"
# Add custom arguments
args.extend([
"--ip=0.0.0.0",
"--port="+str(self.port),
"--SingleUserNotebookApp.default_url=/lab",
"--notebook-dir="+self.notebook_dir,
"--debug",
"--allow-root"
])
return args
async def _get_nodes(self):
k8s.config.load_incluster_config()
async with k8s.client.api_client.ApiClient() as api_client:
core = k8s.client.CoreV1Api(api_client)
nodes = await core.list_node()
return nodes.items
async def _get_pods(self):
k8s.config.load_incluster_config()
async with k8s.client.api_client.ApiClient() as api_client:
core = k8s.client.CoreV1Api(api_client)
pods = await core.list_pod_for_all_namespaces()
return pods.items
@property
def options_form(self):
# Dynamically generate the form
return self.generate_options_form()
def generate_options_form(self):
options_to_return = """
<style>
.memory-button {
padding: 12px 20px;
margin: 2px;
border: none;
border-radius: 8px;
font-size: 1em;
font-weight: bold;
background: linear-gradient(135deg, #4caf50, #81c784);
color: white;
cursor: pointer;
transition: all 0.3s ease-in-out;
box-shadow: 0 4px 6px rgba(0, 0, 0, 0.1);
}
.memory-button:hover {
background: linear-gradient(135deg, #388e3c, #66bb6a);
transform: scale(1.05);
box-shadow: 0 6px 8px rgba(0, 0, 0, 0.15);
}
.memory-button:active {
transform: scale(0.95);
box-shadow: 0 3px 5px rgba(0, 0, 0, 0.2);
}
.memory-button.selected {
background: linear-gradient(135deg, #2e7d32, #43a047);
box-shadow: 0 6px 8px rgba(0, 0, 0, 0.2);
transform: scale(1.05);
}
.cpu-button {
padding: 12px 20px;
margin: 2px;
border: none;
border-radius: 8px;
font-size: 1em;
font-weight: bold;
background: linear-gradient(135deg, #2196f3, #64b5f6); /* Blue Gradient 1 */
color: white;
cursor: pointer;
transition: all 0.3s ease-in-out;
box-shadow: 0 4px 6px rgba(0, 0, 0, 0.1);
}
.cpu-button:hover {
background: linear-gradient(135deg, #1976d2, #42a5f5); /* Blue Gradient 2 */
transform: scale(1.05);
box-shadow: 0 6px 8px rgba(0, 0, 0, 0.15);
}
.cpu-button:active {
transform: scale(0.95);
box-shadow: 0 3px 5px rgba(0, 0, 0, 0.2);
}
.cpu-button.selected {
background: linear-gradient(135deg, #1565c0, #1e88e5); /* Blue Gradient 3 */
box-shadow: 0 6px 8px rgba(0, 0, 0, 0.2);
transform: scale(1.05);
}
</style>
<script>
function setMemory(memorySize, button) {
const selectElement = document.getElementById('mem-select');
selectElement.value = memorySize;
const buttons = document.querySelectorAll('.memory-button');
buttons.forEach(button => button.classList.remove('selected'));
const clickedButton = Array.from(buttons).find(button => button.textContent.includes(memorySize.replace('G', 'GB')));
if (clickedButton) clickedButton.classList.add('selected');
}
function setCpu(cpuSize, button) {
const selectElement = document.getElementById('cpu-select');
selectElement.value = cpuSize;
const buttons = document.querySelectorAll('.cpu-button');
buttons.forEach(button => button.classList.remove('selected'));
const clickedButton = Array.from(buttons).find(button => button.textContent.includes(cpuSize));
if (clickedButton) clickedButton.classList.add('selected');
}
</script>
<label for="stack">Select your desired image:</label>
<br>
<input type="radio" id="option1" name="img" value="ghcr.io/dodas-ts/htc-dask-wn:v1.0.6-ml-infn-ssh-v5">
<label for="option1">ghcr.io/dodas-ts/htc-dask-wn:v1.0.6-ml-infn-ssh-v5</label><br>
<a href="https://github.com/DODAS-TS/dodas-docker-images" target="_blank">Source docker image from DODAS</a>
<br>
<input type="radio" id="option2" name="img" value="biancoj/jlab-ai">
<label for="option2">biancoj/jlab-ai</label><br>
<a href="https://github.com/landerlini/ai-infn-platform/tree/main/docker" target="_blank">Source docker image (from ai-infn platform)</a>
<br>
<input type="radio" id="option3" name="img" value="/cvmfs/datacloud.infn.it/test/jlab-ssh">
<label for="option3">/cvmfs/datacloud.infn.it/test/jlab-ssh</label><br>
<a href="https://github.com/DODAS-TS/dodas-docker-images" target="_blank">Source docker image from DODAS</a>
<br>
<input type="radio" id="option4" name="img" value="/cvmfs/unpacked.infn.it/harbor.cloud.infn.it/unpacked/htc-dask-wn:v1.0.6-ml-infn-ssh-v5">
<label for="option4">/cvmfs/unpacked.infn.it/harbor.cloud.infn.it/unpacked/htc-dask-wn:v1.0.6-ml-infn-ssh-v5</label><br>
<a href="https://github.com/DODAS-TS/dodas-docker-images" target="_blank">Source docker image from DODAS</a>
<br>
<input type="radio" id="option5" name="img" value="jlab-fpga:0.1">
<label for="option5">jlab-fpga:0.1</label><br>
<a href="" target="_blank">Custom FPGA image</a>
<br>
<br>
<div style="text-align: center;">
<label for="cpu" style="font-size: 1em; margin-bottom: 5px; font-weight: bold; display: inline-block;">
CPUs
</label>
<div id="cpu">
<button type="button" onclick="setCpu(1, this)" class="cpu-button">1</button>
<button type="button" onclick="setCpu(2, this)" class="cpu-button">2</button>
<button type="button" onclick="setCpu(4, this)" class="cpu-button">4</button>
<button type="button" onclick="setCpu(8, this)" class="cpu-button">8</button>
</div>
</div>
<select name="cpu" id="cpu-select" size="1" style="display: none;">
<option value="1">1</option>
<option value="2">2</option>
<option value="4">4</option>
<option value="8">8</option>
</select>
<br>
<div style="text-align: center;">
<label for="mem" style="font-size: 1em; margin-bottom: 5px; font-weight: bold; display: inline-block;">
RAM
</label>
<div id="mem">
<button type="button" onclick="setMemory('2G', this)" class="memory-button">2GB</button>
<button type="button" onclick="setMemory('4G', this)" class="memory-button">4GB</button>
<button type="button" onclick="setMemory('8G', this)" class="memory-button">8GB</button>
<button type="button" onclick="setMemory('16G', this)" class="memory-button">16GB</button>
<button type="button" onclick="setMemory('32G', this)" class="memory-button">32GB</button>
<button type="button" onclick="setMemory('64G', this)" class="memory-button">64GB</button>
</div>
</div>
<select name="mem" id="mem-select" size="1" style="display: none;">
<option value="2G">2GB</option>
<option value="4G">4GB</option>
<option value="8G">8GB</option>
<option value="16G">16GB</option>
<option value="32G">32GB</option>
<option value="64G">64GB</option>
</select>
<br>
<br>
"""
nest_asyncio.apply()
nodes = asyncio.run(self._get_nodes())
vk_nodes = [node for node in nodes if node.metadata.labels.get('virtual-node.interlink/type') == 'virtual-kubelet']
nodes_labels = []
accelerator_labels = []
accelerator_labels.append("none")
available_gpus = 0
available_fpags = 0
for node in vk_nodes:
# append the node label to the list nodes_labels
nodes_labels.append({ "hostname": node.metadata.name, "label": node.metadata.labels.get('accelerator', '')})
available_gpus += int(node.status.capacity.get('nvidia.com/gpu', 0))
available_fpags += int(node.status.capacity.get('xilinx.com/fpga', 0))
if any(gpu['model'] == node.metadata.labels.get('accelerator', '') for gpu in self.gpus_model_known):
if node.metadata.labels.get('accelerator', '') not in accelerator_labels:
accelerator_labels.append(node.metadata.labels.get('accelerator', ''))
self.map_node_gpu[node.metadata.labels.get('accelerator', '')] = { "hostname": node.metadata.name, "gpus": int(node.status.capacity.get('nvidia.com/gpu', 0))}
self.gpus_status[node.metadata.labels.get('accelerator', '')] = { 'total': int(node.status.capacity.get('nvidia.com/gpu', 0)), 'used': 0, 'available': int(node.status.capacity.get('nvidia.com/gpu', 0)) }
elif node.metadata.labels.get('accelerator', '') == "U55C":
if node.metadata.labels.get('accelerator', '') not in accelerator_labels:
accelerator_labels.append(node.metadata.labels.get('accelerator', ''))
self.map_node_fpga[node.metadata.labels.get('accelerator', '')] = { "hostname": node.metadata.name, "fpgas": int(node.status.capacity.get('xilinx.com/fpga', 0))}
self.fpga_status[node.metadata.labels.get('accelerator', '')] = { 'total': int(node.status.capacity.get('xilinx.com/fpga', 0)), 'used': 0, 'available': int(node.status.capacity.get('xilinx.com/fpga', 0)) }
elif node.metadata.labels.get('accelerator', '') == "none":
self.map_node_gpu[node.metadata.labels.get('accelerator', '')] = { "hostname": node.metadata.name, "gpus": 0}
already_allocated_gpus = 0
already_allocated_fpgas = 0
pods = asyncio.run(self._get_pods())
running_pods = [pod for pod in pods if pod.status.phase == "Running"]
for pod in running_pods:
try:
for container in pod.spec.containers:
if container.resources and 'nvidia.com/gpu' in container.resources.limits:
# get the label of the node where the pod is running
node_label = pod.spec.node_name
accelerator_of_node = [node["label"] for node in nodes_labels if node["hostname"] == node_label][0]
self.gpus_status[accelerator_of_node]["used"] += int(container.resources.limits['nvidia.com/gpu'])
self.gpus_status[accelerator_of_node]["available"] -= int(container.resources.limits['nvidia.com/gpu'])
already_allocated_gpus += int(container.resources.limits['nvidia.com/gpu'])
elif container.resources and 'xilinx.com/fpga' in container.resources.limits:
# get the label of the node where the pod is running
node_label = pod.spec.node_name
accelerator_of_node = [node["label"] for node in nodes_labels if node["hostname"] == node_label][0]
self.fpga_status[accelerator_of_node]["used"] += int(container.resources.limits['xilinx.com/fpga'])
self.fpga_status[accelerator_of_node]["available"] -= int(container.resources.limits['xilinx.com/fpga'])
already_allocated_fpgas += int(container.resources.limits['xilinx.com/fpga'])
except Exception as e:
pass
if self.gpus_status:
options_to_return += '<table style="width:100%">'
options_to_return += '<tr>'
options_to_return += '<th>GPU Model</th>'
options_to_return += '<th>Total GPUs</th>'
options_to_return += '<th>Used GPUs</th>'
options_to_return += '<th>Available GPUs</th>'
options_to_return += '</tr>'
for key, value in self.gpus_status.items():
# Find the image URL for the current GPU model
image_url = next((gpu['image_url'] for gpu in self.gpus_model_known if gpu['model'] == key), None)
options_to_return += '<tr>'
options_to_return += '<td>'
# Add GPU image if URL is found
if image_url:
options_to_return += f'<img src="{image_url}" alt="{key}" style="width:60px; height:60px; border-radius:50%; vertical-align:middle; margin-right:10px;">'
# Add GPU model name
options_to_return += f'{key}</td>'
options_to_return += f'<td>{value["total"]}</td>'
options_to_return += f'<td>{value["used"]}</td>'
options_to_return += f'<td>{value["available"]}</td>'
options_to_return += '</tr>'
options_to_return += '</table>'
options_to_return += '<br>'
# if available_gpus > 0:
# options_to_return += f"<p>Total GPUs available: <b style='color: darkgreen;'>{available_gpus}</b></p>"
# if already_allocated_gpus > 0:
# options_to_return += f"<p>Used GPUs: <b style='color: darkred;'>{already_allocated_gpus}</b></p>"
# else:
# options_to_return += f"<p>Used GPUs: <b>0</b></p>"
# unused_gpus = available_gpus - already_allocated_gpus
# if unused_gpus > 0:
# options_to_return += f"<p>Unused GPUs: <b>{unused_gpus}</b></p>"
# else:
# options_to_return += f"<p>Unused GPUs: <b>0</b></p>"
if available_gpus > 0:
options_to_return += '<label for="gpu">Select your desired number of GPUs:</label>'
options_to_return += '<select name="gpu" size="1">'
for i in range(0, int(available_gpus)+1):
options_to_return += f'<option value="{i}">{i}</option>'
options_to_return += "</select><br>"
options_to_return += "<br>"
if self.fpga_status:
options_to_return += '<table style="width:100%">'
options_to_return += '<tr>'
options_to_return += '<th>FPGA Model</th>'
options_to_return += '<th>Total FPGAs</th>'
options_to_return += '<th>Used FPGAs</th>'
options_to_return += '<th>Available FPGAs</th>'
options_to_return += '</tr>'
for key, value in self.fpga_status.items():
image_url = next((fpga['image_url'] for fpga in self.fpgas_model_known if fpga['model'] == key), None)
options_to_return += '<tr>'
options_to_return += '<td>'
if image_url:
options_to_return += f'<img src="{image_url}" alt="{key}" style="width:60px; height:60px; border-radius:50%; vertical-align:middle; margin-right:10px;">'
options_to_return += f'{key}</td>'
options_to_return += f'<td>{value["total"]}</td>'
options_to_return += f'<td>{value["used"]}</td>'
options_to_return += f'<td>{value["available"]}</td>'
options_to_return += '</tr>'
options_to_return += '</table>'
options_to_return += '<br>'
# if available_fpags > 0:
# options_to_return += f"<p>Total FPGAs available: <b style='color: darkgreen;'>{available_fpags}</b></p>"
# if already_allocated_fpgas > 0:
# options_to_return += f"<p>Used FPGAs: <b style='color: darkred;'>{already_allocated_fpgas}</b></p>"
# else:
# options_to_return += f"<p>Used FPGAs: <b>0</b></p>"
# unused_fpgas = available_fpags - already_allocated_fpgas
# if unused_fpgas > 0:
# options_to_return += f"<p>Unused FPGAs: <b>{unused_fpgas}</b></p>"
# else:
# options_to_return += f"<p>Unused FPGAs: <b>0</b></p>"
options_to_return += '<label for="fpga">Select your desired number of FPGAs:</label>'
options_to_return += '<select name="fpga" size="1">'
for i in range(0, int(available_fpags)+1):
options_to_return += f'<option value="{i}">{i}</option>'
options_to_return += "</select><br>"
options_to_return += '<label for="offload">Enable Offloading to:</label>'
options_to_return += '<select name="offload" size="1">'
for label in accelerator_labels:
options_to_return += f'<option value="{label}">{label}</option>'
# if already_allocated_gpus > 0:
# options_to_return += '<p><b>You cannot use a GPU because someone else is using it</b></p>'
# if already_allocated_fpgas > 0:
# options_to_return += '<p><b>You cannot use a FPGA because someone else is using it</b></p>'
options_to_return += "</select><br>"
return options_to_return
def options_from_form(self, formdata):
options = {}
options['img'] = formdata['img']
container_image = ''.join(formdata['img'])
self.image = container_image
options['cpu'] = formdata['cpu']
cpu = ''.join(formdata['cpu'])
self.cpu_guarantee = float(cpu)
self.cpu_limit = float(cpu)
options['mem'] = formdata['mem']
memory = ''.join(formdata['mem'])
self.mem_guarantee = memory
self.mem_limit = memory
options['offload'] = ''.join(formdata['offload'])
if 'gpu' in formdata:
options['gpu'] = formdata['gpu']
gpu = ''.join(formdata['gpu'])
if 'fpga' in formdata:
options['fpga'] = formdata['fpga']
fpga = ''.join(formdata['fpga'])
sock = socket.socket()
sock.bind(('', 0))
self.port = sock.getsockname()[1]
if options['offload'] == 'NO' or options['offload'] == 'none':
self.tolerations = [
{
"key": "accelerator",
"operator": "Equal",
"value": "none",
"effect": "NoSchedule"
},
# {
# "key": "virtual-node.interlink/no-schedule=false",
# "value": "false",
# "effect": "NoSchedule"
# }
]
elif options['offload'] in [gpu['model'] for gpu in self.gpus_model_known]:
self.tolerations = [
{
"key": "accelerator",
"operator": "Equal",
"value": options['offload'],
"effect": "NoSchedule"
},
# {
# "key": "virtual-node.interlink/no-schedule=false",
# "value": "false",
# "effect": "NoSchedule"
# }
]
self.extra_resource_guarantees = {"nvidia.com/gpu": gpu}
self.extra_resource_limits = {"nvidia.com/gpu": gpu}
elif options['offload'] == 'U55C':
self.tolerations = [
{
"key": "accelerator",
"operator": "Equal",
"value": options['offload'],
"effect": "NoSchedule"
},
# {
# "key": "virtual-node.interlink/no-schedule",
# "value": "false",
# "effect": "NoSchedule"
# }
]
self.extra_resource_guarantees = {"xilinx.com/fpga": fpga}
self.extra_resource_limits = {"xilinx.com/fpga": fpga}
if 'poc' in options['offload']:
pre_exec_value = f'afuse_cvmfs2_helper && ls /cvmfs/datacloud.infn.it && mkdir -p notebooks/{self.user.name}'
flags_value = f'-t 100 -A inf24_lhc_1 --gres=gpu:{gpu} --reservation=test_cvmfs -p boost_usr_prod -w lrdn0241'
singularity_options = f'--no-home --bind notebooks/{self.user.name}:/home/{self.user.name}'
self.extra_annotations = {"job.vk.io/pre-exec": pre_exec_value,
"slurm-job.vk.io/flags": flags_value,
"slurm-job.vk.io/singularity-options": singularity_options}
if 'unpacked' in self.image:
self.extra_annotations.update({"job.vk.io/pre-exec": f'export SINGULARITY_USERNS=1 && mkdir -p notebooks/{self.user.name} && afuse_cvmfs2_helper && ls /cvmfs/datacloud.infn.it'})
self.services_enabled = True
self.extra_labels = { "app": "jupyterhub", "component": "hub", "release": "helm-jhub-release"}
if self.image == "biancoj/jlab-ai":
self.notebook_dir = "/home/jovyan"
else:
self.notebook_dir = "/jupyter-workspace"
#self.notebook_dir = f'/home/{self.user.name}'
return options
def get_service_manifest(self, owner_reference):
"""
Make a service manifest for dns.
"""
labels = self._build_common_labels(self._expand_all(self.extra_labels))
annotations = self._build_common_annotations(
self._expand_all(self.extra_annotations)
)
from kubernetes_asyncio.client.models import ( V1ObjectMeta, V1Service, V1ServiceSpec, V1ServicePort)
metadata = V1ObjectMeta(
name=self.pod_name,
annotations=annotations,
labels=labels,
owner_references=[owner_reference],
)
service = V1Service(
kind='Service',
metadata=metadata,
spec=V1ServiceSpec(
type='ClusterIP',
ports=[V1ServicePort(name='http', port=self.port, target_port=self.port)],
selector={ "app": "jupyterhub", "component": "hub", "release": "helm-jhub-release"}
),
)
return service
async def custom_function(self):
print("Running custom function before starting the notebook server")
await asyncio.sleep(1) # Simulating some async operation
async def start(self):
# Run your custom function here
await self.custom_function()
# Call the parent class's start method to actually start the notebook
return await super().start()
@property
def environment(self):
# dciangot: create an ssh connection on a random port
environment = {
"JUPYTERHUB_SINGLEUSER_EXTENSION": "0",
"JHUB_HOST": jhub_host,
"SSH_PORT": "31022",
"FWD_PORT": f"{self.port}",
"JUPYTERHUB_API_URL": jhub_api_url,
"JUPYTERHUB_ACTIVITY_URL": f"{jhub_api_url}/users/{self.user.name}/activity",
"JUPYTERHUB_SERVICE_URL": f"http://0.0.0.0:{self.port}",
"JUPYTERHUB_SERVER_NAME": "development",
"JUPYTERHUB_HOST": f"https://{jhub_host}:{jhub_port}",
#"JUPYTERHUB_OAUTH_ACCESS_SCOPES": "none", # to understand if it is strictly necessary for slurm plugin to set this to none
#"JUPYTERHUB_OAUTH_SCOPES": "none" # # to understand if it is strictly necessary for slurm plugin to set this to none
}
if 'poc' in self.image:
environment.update({"JUPYTERHUB_OAUTH_ACCESS_SCOPES": "none", "JUPYTERHUB_OAUTH_SCOPES": "none"})
return environment
@property
def node_selector(self):
node_selector = { "kubernetes.io/role": "agent",
"beta.kubernetes.io/os": "linux",
"virtual-node.interlink/type" : "virtual-kubelet"}
if self.user_options.get('offload') in [gpu['model'] for gpu in self.gpus_model_known]:
node_selector.update({"kubernetes.io/hostname" : self.map_node_gpu[self.user_options.get('offload')]["hostname"]})
elif self.user_options.get('offload')=="U55C":
node_selector.update({"kubernetes.io/hostname" : self.map_node_fpga[self.user_options.get('offload')]["hostname"]})
if self.user_options.get('offload')=="N":
node_selector = {}
return node_selector
@property
def volume_mounts(self):
return [
{
'name': f'{self.user.name}-volume',
'mountPath': self.notebook_dir
},
]
@property
def volumes(self):
return [
{
'name': f'{self.user.name}-volume',
'hostPath': {
'path': f'/home/workspace/persistent-storage/{self.user.name}',
'type': 'DirectoryOrCreate',
},
},
]
c.JupyterHub.spawner_class = CustomSpawner
c.KubeSpawner.cmd = [" "]
c.KubeSpawner.args = [" "]
c.KubeSpawner.delete_stopped_pods = False
c.KubeSpawner.privileged = True
c.KubeSpawner.allow_privilege_escalation = True
c.KubeSpawner.extra_pod_config = {
"automountServiceAccountToken": True,
}
c.KubeSpawner.init_containers = []
c.KubeSpawner.debug = True
c.KubeSpawner.services_enabled = True
c.KubeSpawner.extra_labels = { "app": "jupyterhub", "component": "hub", "release": "helm-jhub-release"}
c.KubeSpawner.extra_container_config = {
"securityContext": {
"privileged": True,
"capabilities": {
"add": ["SYS_ADMIN"]
}
}
}
c.KubeSpawner.http_timeout = 60
c.KubeSpawner.start_timeout = 60
#c.KubeSpawner.notebook_dir = "/home/jovyan"