-
Notifications
You must be signed in to change notification settings - Fork 2
/
Copy pathfuzzball.txt
81 lines (49 loc) · 2.32 KB
/
fuzzball.txt
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
ciq
k8s-dev-1
fuzzball context # can suport multiple fuzzball cluster, and change via context. # think lrc vs brc
# recorded 12:22pm 1.18
fuzzall context create default api....nip..io https:// Fuzzball
fuzzball context list
fz context login
# on a separate terminal, login via browser (text browser w3m can work, albeit bit clunky)
w3m # text based browser
w3m https://auth.192.168...
1 # to accept self-signed cert
login as fuzzball-admin
workload, pipeline, like job that runs in hpc cluster
fz workflow
fz workflow start # use yaml file
# can also use template, client or server side
# pseudo cli interface
have 3 control plane node and 1 worker node in the dev env
container will be started with uid 0
use ldap to map known user id
port-forward... eg used to run jupyther notebook on compute node and have browser usable from a client
# utilize geopc api for this?
# user don't need to ssh to anything
# mostly a gui
# install cli (on laptop) and use that cli for all work (no need to ssh to server, that's the design goal)
# John concern about NFS port forward, below 1024, then get send to client... don't want client to NFS mount from using fuzzball cli/api
DSL - some job descrition language
define the workflow
12:38 list sample dsl
container, can do docker, github registry, can do private with username/pass
fz workflow start --name printer-test printer.yaml # 12:46pm
fz workflow status --watch --name printer-test hexid
fz workflow log -f -name printer-test hexid
# -f = follow, grap std out/err and display them (with some mark up)
## the hexid is returned once a workflow is returned. like docker instance id?
## IMHO, this hexid is really annoying. i submit 10 jobs and want to check their status. now have to find all such hexid, rather than have 10 sequential job id.
fz schedule hosts
${{ }} server side template # 12:55pm
%{{ }} client side template
${{ secrtet "users/user/aws/AccessKeyId" }}
${{ secrtet "users/user/aws/AccessKeySecret" }}
egres:
move data out of the volume created
fz workflow exec --ty --namme printer-test hexid printer /bin/sh
# so really just wrapper around docker
# but why they need both name and hexid? so redundant
# container run as root
# not in username space
# integration with ldap... run under correct context (inside the container), need id provier for this to work