[cherry-pick][doc][train] Clarify error message when trying to use local storage for multi-node distributed training and checkpointing #41844
Conversation
This file contains 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
justinvyu
requested review from
richardliaw,
krfricke,
xwjiang2010,
amogkam,
matthewdeng,
Yard1,
maxpumperla,
woshiyyya and
a team
as code owners
zhe-thoughts
approved these changes
Dec 12, 2023
|
Doc build failed, restarting tests |
architkulkarni
added
release-blocker
8 tasks
…or multi-node distributed training and checkpointing (ray-project#41832) Ray 2.7 removed support for using the head node as the persistent storage for checkpoints and artifacts in a multi-node distributed training. The alternative recommendation is to use cloud storage or a shared filesystem instead via `RunConfig(storage_path)`. Ray Train/Tune will error if the user attempts to checkpoint `ray.train.report(..., checkpoint=...)` from a worker that's on a remote node. This is because the new assumption is that all worker nodes have access to read/write from the same persistent storage, and the "head node local storage" is not accessible by all nodes. However, the error message that shows up is confusing. All nodes can technically access the local path in the message -- the problem is that not all nodes can access the the SAME local path. This PR improves the error message to make this more clear and to suggest an actionable fix. This PR also updates most of the getting started user guides to mention the multi-node storage requirement and links to the storage user guide. --------- Signed-off-by: Justin Yu <justinvyu@anyscale.com>
architkulkarni
merged commit 0c5a3ec
into
ray-project:releases/2.9.0
15 of 16 checks passed
justinvyu
removed
the
@author-action-required
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.
Why are these changes needed?
Related issue number
This is a cherry-pick of #41832. This has 2 commits because the 2nd pick has many conflicts if merged without the first docs change PR. Both of these are primarily doc changes.
Checks
git commit -s) in this PR.scripts/format.shto lint the changes in this PR.method in Tune, I've added it in
doc/source/tune/api/under thecorresponding
.rstfile.