Default interfaces and other things #9
Merged
Conversation
This file contains hidden or bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
Codecov ReportAttention: Patch coverage is
Additional details and impacted files@@ Coverage Diff @@
## master #9 +/- ##
==========================================
+ Coverage 86.37% 86.83% +0.45%
==========================================
Files 10 11 +1
Lines 1982 2051 +69
==========================================
+ Hits 1712 1781 +69
Misses 270 270 ☔ View full report in Codecov by Sentry. |
jpsamaroo
approved these changes
Nov 25, 2024
- Make LocalProcess.bind_port an Int so it doesn't print in hex. - Use @error when launching workers so we can show the full backtrace if it fails.
`Base.active_project()` is the public API for this.
It's otherwise very annoying and error-prone to set these with every call to `addprocs()`.
JamesWrigley
force-pushed
the
default-interface
branch
from
f4e0443 to
915eb61
Compare
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
Add this suggestion to a batch that can be applied as a single commit.
This suggestion is invalid because no changes were made to the code.
Suggestions cannot be applied while the pull request is closed.
Suggestions cannot be applied while viewing a subset of changes.
Only one suggestion per line can be applied in a batch.
Add this suggestion to a batch that can be applied as a single commit.
Applying suggestions on deleted lines is not supported.
You must change the existing code in this line in order to create a valid suggestion.
Outdated suggestions cannot be applied.
This suggestion has been applied or marked resolved.
Suggestions cannot be applied from pending reviews.
Suggestions cannot be applied on multi-line comments.
Suggestions cannot be applied while the pull request is queued to merge.
Suggestion cannot be applied right now. Please check back later.
This PR has a few changes, but the important ones are:
Default worker interface
We now try to automatically pick the fastest interface for the worker to listen on. This is done by using a minimal version of NetworkInterfaceControllers.jl (thanks @JBlaschke ❤️) and a Linux-only check for the negotiated link speed. In the future we could add support for passing an interface name/CIDR blocks/interface type to
addprocs()for explicit selection.Unfortunately testing the functionality is a bit tricky, for now there's just a sanity test in the tests and I've tested it manually on two clusters with both ethernet and infiniband networks.
Passing environment variables
addprocs(::SSHManager)now passes allJULIA_*environment variables by default. I found out we didn't do this already when I added a worker on a node with a different microarchitecture and saw segfaults on the local node from the pkgimages being compiled without all the archs inJULIA_CPU_TARGET... For context, these are the environment variables our Julia module sets by default:All of them are important for a good out-of-the-box experience. On the other hand, admittedly this change is very much biased towards adding workers on a cluster with a shared filesystem. Personally I think it's a good default, but alternatively we could support a
JULIA_DISTRIBUTEDNEXT_PASS_ALL_VARSenv var for admins to set that will enable/disable passing everything by default.