-
Notifications
You must be signed in to change notification settings - Fork 96
New issue
Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.
By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.
Already on GitHub? Sign in to your account
Fix UCX examples for InfiniBand #556
Fix UCX examples for InfiniBand #556
Conversation
Codecov Report
@@ Coverage Diff @@
## branch-0.19 #556 +/- ##
================================================
+ Coverage 61.06% 92.46% +31.39%
================================================
Files 22 16 -6
Lines 2571 1605 -966
================================================
- Hits 1570 1484 -86
+ Misses 1001 121 -880
Continue to review full report at Codecov.
|
examples/ucx/client_initialize.py
Outdated
x = rs.random((10000, 10000), chunks=1000) | ||
x.sum().compute() | ||
|
||
# shutdown client |
There was a problem hiding this comment.
Choose a reason for hiding this comment
The reason will be displayed to describe this comment to others. Learn more.
# shutdown client | |
# shutdown cluster |
There was a problem hiding this comment.
Choose a reason for hiding this comment
The reason will be displayed to describe this comment to others. Learn more.
This shuts down the entire cluster, not the client: https://distributed.dask.org/en/latest/api.html#distributed.Client.shutdown .
examples/ucx/client_initialize.py
Outdated
client = Client(address) # noqa F841 | ||
client = Client(address) | ||
|
||
# client code here |
There was a problem hiding this comment.
Choose a reason for hiding this comment
The reason will be displayed to describe this comment to others. Learn more.
This is a bit misleading, I would rename that to user code. Saying client code may sound like it's executed on the client-side only.
examples/ucx/local_cuda_cluster.py
Outdated
x = rs.random((10000, 10000), chunks=1000) | ||
x.sum().compute() | ||
|
||
# shutdown client |
There was a problem hiding this comment.
Choose a reason for hiding this comment
The reason will be displayed to describe this comment to others. Learn more.
# shutdown client | |
# shutdown cluster |
Thanks for the review @pentschev! |
There was a problem hiding this comment.
Choose a reason for hiding this comment
The reason will be displayed to describe this comment to others. Learn more.
LGTM, thanks @charlesbluca !
@gpucibot merge |
Some small changes to the UCX examples to fix some IB-related issues I discussed with @pentschev:
--scheduler-file
to linkdask-scheduler
todask-cuda-workers
, aslocalhost
cannot be used when IB is enabled