fix: checkpoint workload fails if upload fails #752
Merged
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
stoksc
approved these changes
Jun 22, 2020
| @@ -0,0 +1,107 @@ | |||
| import contextlib | |||
There was a problem hiding this comment.
praise: tests!
We talked a bit about moving this into a test utility file offline, decided this stuff is all workload manager specific for the most part.
| checkpoint_info.get("framework", ""), | ||
| checkpoint_info.get("format", ""), | ||
| ) | ||
| if self.rendezvous_info.get_rank() != 0: |
There was a problem hiding this comment.
GitHub diffs suck 🙃, but I like losing the indentation.
Previously, the ordering of the context manager for storage_manager.save_path() was such that workload completed messages were sent before the upload finished. As a result, the database would sometimes contain checkpoint uuid's that never got uploaded to cloud storage.
rb-determined-ai
force-pushed
the
upload-failure
branch
from
7c91cdb
to
4ae3526
Compare
tayritenour
pushed a commit
to tayritenour/determined
that referenced
this pull request
Apr 25, 2023
…etermined-ai#752) Deletion of the dispatch is done synchronously in the DispatchExited event handler, need to make it async to avoid blocking the event handler. Moved content of DispatchExited to a go routine (dispatchExited) except for accesses to m.reqList which should remain to avoid the need for additional synchronization. Identified one other synchronous call to m.removeDispatchEnvironment that needed to be made async. Added comments to other call sites indicating they are already invoked from an existing go routine so are non-blocking. Extracted the m.reqList use from startLauncherJob go routine back into the event handler to avoid the need for additional synchronization.
eecsliu
pushed a commit
to eecsliu/determined
that referenced
this pull request
Jun 23, 2023
…etermined-ai#752) Deletion of the dispatch is done synchronously in the DispatchExited event handler, need to make it async to avoid blocking the event handler. Moved content of DispatchExited to a go routine (dispatchExited) except for accesses to m.reqList which should remain to avoid the need for additional synchronization. Identified one other synchronous call to m.removeDispatchEnvironment that needed to be made async. Added comments to other call sites indicating they are already invoked from an existing go routine so are non-blocking. Extracted the m.reqList use from startLauncherJob go routine back into the event handler to avoid the need for additional synchronization.
stoksc
pushed a commit
that referenced
this pull request
Jun 26, 2023
…752) Deletion of the dispatch is done synchronously in the DispatchExited event handler, need to make it async to avoid blocking the event handler. Moved content of DispatchExited to a go routine (dispatchExited) except for accesses to m.reqList which should remain to avoid the need for additional synchronization. Identified one other synchronous call to m.removeDispatchEnvironment that needed to be made async. Added comments to other call sites indicating they are already invoked from an existing go routine so are non-blocking. Extracted the m.reqList use from startLauncherJob go routine back into the event handler to avoid the need for additional synchronization.
eecsliu
pushed a commit
that referenced
this pull request
Jun 28, 2023
…752) Deletion of the dispatch is done synchronously in the DispatchExited event handler, need to make it async to avoid blocking the event handler. Moved content of DispatchExited to a go routine (dispatchExited) except for accesses to m.reqList which should remain to avoid the need for additional synchronization. Identified one other synchronous call to m.removeDispatchEnvironment that needed to be made async. Added comments to other call sites indicating they are already invoked from an existing go routine so are non-blocking. Extracted the m.reqList use from startLauncherJob go routine back into the event handler to avoid the need for additional synchronization.
eecsliu
pushed a commit
that referenced
this pull request
Jun 28, 2023
…752) Deletion of the dispatch is done synchronously in the DispatchExited event handler, need to make it async to avoid blocking the event handler. Moved content of DispatchExited to a go routine (dispatchExited) except for accesses to m.reqList which should remain to avoid the need for additional synchronization. Identified one other synchronous call to m.removeDispatchEnvironment that needed to be made async. Added comments to other call sites indicating they are already invoked from an existing go routine so are non-blocking. Extracted the m.reqList use from startLauncherJob go routine back into the event handler to avoid the need for additional synchronization.
stoksc
pushed a commit
that referenced
this pull request
Jul 20, 2023
…752) Deletion of the dispatch is done synchronously in the DispatchExited event handler, need to make it async to avoid blocking the event handler. Moved content of DispatchExited to a go routine (dispatchExited) except for accesses to m.reqList which should remain to avoid the need for additional synchronization. Identified one other synchronous call to m.removeDispatchEnvironment that needed to be made async. Added comments to other call sites indicating they are already invoked from an existing go routine so are non-blocking. Extracted the m.reqList use from startLauncherJob go routine back into the event handler to avoid the need for additional synchronization.
eecsliu
pushed a commit
that referenced
this pull request
Jul 24, 2023
…752) Deletion of the dispatch is done synchronously in the DispatchExited event handler, need to make it async to avoid blocking the event handler. Moved content of DispatchExited to a go routine (dispatchExited) except for accesses to m.reqList which should remain to avoid the need for additional synchronization. Identified one other synchronous call to m.removeDispatchEnvironment that needed to be made async. Added comments to other call sites indicating they are already invoked from an existing go routine so are non-blocking. Extracted the m.reqList use from startLauncherJob go routine back into the event handler to avoid the need for additional synchronization.
stoksc
pushed a commit
that referenced
this pull request
Oct 17, 2023
…752) Deletion of the dispatch is done synchronously in the DispatchExited event handler, need to make it async to avoid blocking the event handler. Moved content of DispatchExited to a go routine (dispatchExited) except for accesses to m.reqList which should remain to avoid the need for additional synchronization. Identified one other synchronous call to m.removeDispatchEnvironment that needed to be made async. Added comments to other call sites indicating they are already invoked from an existing go routine so are non-blocking. Extracted the m.reqList use from startLauncherJob go routine back into the event handler to avoid the need for additional synchronization.
azhou-determined
pushed a commit
that referenced
this pull request
Dec 7, 2023
…752) Deletion of the dispatch is done synchronously in the DispatchExited event handler, need to make it async to avoid blocking the event handler. Moved content of DispatchExited to a go routine (dispatchExited) except for accesses to m.reqList which should remain to avoid the need for additional synchronization. Identified one other synchronous call to m.removeDispatchEnvironment that needed to be made async. Added comments to other call sites indicating they are already invoked from an existing go routine so are non-blocking. Extracted the m.reqList use from startLauncherJob go routine back into the event handler to avoid the need for additional synchronization.
wes-turner
pushed a commit
that referenced
this pull request
Feb 2, 2024
…752) Deletion of the dispatch is done synchronously in the DispatchExited event handler, need to make it async to avoid blocking the event handler. Moved content of DispatchExited to a go routine (dispatchExited) except for accesses to m.reqList which should remain to avoid the need for additional synchronization. Identified one other synchronous call to m.removeDispatchEnvironment that needed to be made async. Added comments to other call sites indicating they are already invoked from an existing go routine so are non-blocking. Extracted the m.reqList use from startLauncherJob go routine back into the event handler to avoid the need for additional synchronization.
rb-determined-ai
pushed a commit
that referenced
this pull request
Feb 29, 2024
…752) Deletion of the dispatch is done synchronously in the DispatchExited event handler, need to make it async to avoid blocking the event handler. Moved content of DispatchExited to a go routine (dispatchExited) except for accesses to m.reqList which should remain to avoid the need for additional synchronization. Identified one other synchronous call to m.removeDispatchEnvironment that needed to be made async. Added comments to other call sites indicating they are already invoked from an existing go routine so are non-blocking. Extracted the m.reqList use from startLauncherJob go routine back into the event handler to avoid the need for additional synchronization.
amandavialva01
pushed a commit
that referenced
this pull request
Mar 18, 2024
…752) Deletion of the dispatch is done synchronously in the DispatchExited event handler, need to make it async to avoid blocking the event handler. Moved content of DispatchExited to a go routine (dispatchExited) except for accesses to m.reqList which should remain to avoid the need for additional synchronization. Identified one other synchronous call to m.removeDispatchEnvironment that needed to be made async. Added comments to other call sites indicating they are already invoked from an existing go routine so are non-blocking. Extracted the m.reqList use from startLauncherJob go routine back into the event handler to avoid the need for additional synchronization.
eecsliu
pushed a commit
that referenced
this pull request
Apr 18, 2024
…752) Deletion of the dispatch is done synchronously in the DispatchExited event handler, need to make it async to avoid blocking the event handler. Moved content of DispatchExited to a go routine (dispatchExited) except for accesses to m.reqList which should remain to avoid the need for additional synchronization. Identified one other synchronous call to m.removeDispatchEnvironment that needed to be made async. Added comments to other call sites indicating they are already invoked from an existing go routine so are non-blocking. Extracted the m.reqList use from startLauncherJob go routine back into the event handler to avoid the need for additional synchronization.
eecsliu
pushed a commit
to determined-ai/determined-release-testing
that referenced
this pull request
Apr 22, 2024
…etermined-ai#752) Deletion of the dispatch is done synchronously in the DispatchExited event handler, need to make it async to avoid blocking the event handler. Moved content of DispatchExited to a go routine (dispatchExited) except for accesses to m.reqList which should remain to avoid the need for additional synchronization. Identified one other synchronous call to m.removeDispatchEnvironment that needed to be made async. Added comments to other call sites indicating they are already invoked from an existing go routine so are non-blocking. Extracted the m.reqList use from startLauncherJob go routine back into the event handler to avoid the need for additional synchronization.
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.
Description
Fix a bug that @liamcli found. Asking for a review from @stoksc since it conflicts with a change he has in flight.
Previously, the ordering of the context manager for storage_manager.save_path() was such that workload completed messages were sent before the upload finished. As a result, the database would sometimes contain checkpoint uuid's that never got uploaded to cloud storage.
Test Plan
I added a unit test for the workload manager, which fails without the fix.