fix(test): cleanup snapshot files better #5126
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
The restore_from_snapshot function did not integrate well with uffd-based snapshot restore: Even if a UFFD path was specified, it still created a copy of the snapshot memory file inside the chroot, even though the UFFD handler set this up long ago in space_pf_handler. Fix this, and while we're at it, also remove the need for passing in uffd handler and snapshot file explicitly when using uffd-based restore, as the spawn_pf_handler sets the uffd_handler field of the microvm object, and can also easily be made to actually contain the snapshot from which page faults are being served. Signed-off-by: Patrick Roy <[email protected]>
Codecov ReportAll modified and coverable lines are covered by tests ✅
Additional details and impacted files@@ Coverage Diff @@
## main #5126 +/- ##
=======================================
Coverage 83.09% 83.09%
=======================================
Files 250 250
Lines 26920 26920
=======================================
Hits 22368 22368
Misses 4552 4552
Flags with carried forward coverage won't be shown. Click here to find out more. ☔ View full report in Codecov by Sentry. 🚀 New features to boost your workflow:
|
roypat
force-pushed
the
snapshot-test-cleanup
branch
from
565afda to
eca928f
Compare
roypat
added
the
Status: Awaiting review
pb8o
previously approved these changes
Apr 1, 2025
roypat
force-pushed
the
snapshot-test-cleanup
branch
from
eca928f to
698edb2
Compare
If build_n_from_snapshot is asked to build many snapshots incrementally, it doesn't clean up after itself properly: Each iteration ends by creating a new snapshot of the VM, and this snapshot is passed to iteration n+1. Here, a copy is created inside the new VMs chroot, and iteration n+1 ends by creating a new snapshot for iteration n+2, and deleting the copy of the snapshot inside the chroot. However, we never delete the snapshot created in iteration n, and so with each iteration more snapshots accumulate. Fix this by having the function delete the snapshot created in iteration n after iteration n+2 finished successfully. The idea here is that in case of failure, we will have the snapshot created in iteration n+1 (the one which caused a failure in n+2), and also the snapshot created in n (which was the last known snapshot to successfully go through a test iteration, namely iteration n+1). Signed-off-by: Patrick Roy <[email protected]>
roypat
force-pushed
the
snapshot-test-cleanup
branch
from
698edb2 to
b5652f5
Compare
|
force pushed to fix a typo and a left cover commented debug statement from testing |
kalyazin
approved these changes
Apr 1, 2025
zulinx86
approved these changes
Apr 1, 2025
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
4 participants
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.
It turns out that some of our snapshot tests are not exactly exemplary at cleaning up snapshot memory files after they're done. This is partly because Microvm.restore_from_snapshot was largely oblivious to the fact that if a snapshot is uffd-restored, it doesn't need to be copied into the jail again for mmaping that will never happen, and also due to me messing up the build_n_from_snapshot function and not cleaning up snapshots in "incremental" mode.
License Acceptance
By submitting this pull request, I confirm that my contribution is made under
the terms of the Apache 2.0 license. For more information on following Developer
Certificate of Origin and signing off your commits, please check
CONTRIBUTING.md.PR Checklist
tools/devtool checkstyleto verify that the PR passes theautomated style checks.
how they are solving the problem in a clear and encompassing way.
in the PR.
CHANGELOG.md.Runbook for Firecracker API changes.
integration tests.
TODO.rust-vmm.