-
Notifications
You must be signed in to change notification settings - Fork 0
/
Copy pathautoscaling_config.yaml
152 lines (141 loc) · 4.82 KB
/
autoscaling_config.yaml
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
scaling:
portal_scaling_link: https://simplevm.denbi.de/portal/api/autoscaling
portal_webapp_link: https://simplevm.denbi.de/portal/webapp/#/clusters/overview
scheduler: slurm
active_mode: basic
automatic_update: True
database_reset: False
pattern_id: ''
history_recall: 7
ignore_workers: [ ]
pending_jobs_percent: 1.0
mode:
basic:
info: 'start multiple flavors for 60% pending jobs, no limitations, no time forecast'
service_frequency: 60
limit_memory: 0
limit_worker_starts: 0
limit_workers: 0
limit_flavor_usage:
# 'de.NBI highmem xlarge': 16
# 'de.NBI highmem large': 20
scale_force: 0.6
scale_delay: 60
worker_cool_down: 60
worker_weight: 0
smoothing_coefficient: 0.00
forecast_by_flavor_history: False
job_match_value: 0.95
job_time_threshold: 0.5
job_name_remove_numbers: True
job_name_remove_num_brackets: True
job_name_remove_pattern: ''
job_name_remove_text_within_parentheses: True
forecast_by_job_history: False
forecast_active_worker: 0
forecast_occupied_worker: False
flavor_restriction: 0
flavor_default: ""
flavor_ephemeral: True
flavor_gpu: 1
flavor_depth: -1
large_flavors: False
large_flavors_except_hmf: True
auto_activate_large_flavors: 10
drain_large_nodes: False
drain_only_hmf: False
drain_delay: 0
approach:
info: 'start workers for 60% pending jobs, approach by 10 workers per flavor and service frequency'
limit_worker_starts: 10
scale_force: 0.6
adaptive:
info: 'forecast job time, start worker one flavor ahead in queue, set max 50% hmf worker per service request to drain'
limit_worker_starts: 10
smoothing_coefficient: 0.05
forecast_by_flavor_history: True
forecast_by_job_history: True
forecast_active_worker: 1
job_time_threshold: 0.50
flavor_depth: 1
drain_large_nodes: True
drain_only_hmf: True
drain_delay: -1
sequence:
info: 'forecast job time, start worker for the next jobs in queue with the same flavor'
forecast_by_flavor_history: True
forecast_by_job_history: True
forecast_active_worker: 1
job_time_threshold: 0.50
multi:
info: 'forecast job time, start multiple flavors, automatic higher (threshold 10)'
smoothing_coefficient: 0.01
forecast_by_flavor_history: True
forecast_by_job_history: True
forecast_active_worker: 1
job_time_threshold: 0.50
flavor_depth: -1
large_flavors: False
large_flavors_except_hmf: True
auto_activate_large_flavors: 10
max:
info: 'forecast job time, maximum worker - no flavor separation'
worker_cool_down: 60
worker_weight: 0.00
smoothing_coefficient: 0.00
forecast_by_flavor_history: True
forecast_by_job_history: True
forecast_active_worker: 1
job_time_threshold: 0.50
flavor_restriction: 0.5
flavor_depth: -2
default:
info: 'forecast job time, start multiple flavors, prefer higher flavors (except high memory), smooth time'
scale_delay: 100
worker_cool_down: 70
smoothing_coefficient: 0.01
forecast_by_flavor_history: False
forecast_by_job_history: True
forecast_active_worker: 1
forecast_occupied_worker: True
job_time_threshold: 0.50
flavor_depth: -1
large_flavors: True
large_flavors_except_hmf: True
flavor:
info: 'forecast job time only on flavor level, start multiple flavors, prefer higher flavors (except high memory)'
limit_worker_starts: 10
smoothing_coefficient: 0.00
forecast_by_flavor_history: True
forecast_by_job_history: False
forecast_active_worker: 1
job_time_threshold: 0.4
flavor_depth: 0
large_flavors: False
large_flavors_except_hmf: True
auto_activate_large_flavors: 10
min:
info: 'forecast job time, start a single flavor, aggressive worker drain and scale-down'
limit_worker_starts: 10
worker_cool_down: 0
worker_weight: 0.1
smoothing_coefficient: 0.00
forecast_by_flavor_history: True
forecast_by_job_history: True
forecast_active_worker: 1
job_time_threshold: 0.6
flavor_depth: 0
drain_large_nodes: True
drain_only_hmf: False
drain_delay: 0
reactive:
info: 'forecast job time, start multiple flavors, prefer higher flavors (except high memory) + more workers per runtime + without active worker forecast'
scale_force: 0.8
smoothing_coefficient: 0.02
forecast_by_flavor_history: False
forecast_by_job_history: True
forecast_active_worker: 0
job_time_threshold: 0.20
flavor_depth: -1
large_flavors: True
large_flavors_except_hmf: True