Skip to content
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

Backport fixes for release/2.7 #1775

Merged
merged 4 commits into from
Nov 27, 2024
Merged

Conversation

tiagolobocastro
Copy link
Contributor

test(bdd): make nvme controller usage more robust

Caters for when the device is /dev/nvmeX but X not the same as the controller!

Signed-off-by: Tiago Castro <[email protected]>

Merge #1755

1755: Reuse Rebuild IO handles r=tiagolobocastro a=tiagolobocastro

    fix(rebuild): reuse rebuild IO handles

    Reuses the rebuild IO handles, rather than attempting to allocate
    them per rebuild task.
    The main issue with handle allocation on the fly is that the target
    may have not cleaned up a previous IO qpair connection, and so the
    connect may fail. We started seeing this more on CI because we forgot
    to cherry-pick a commit increasing the retry delay.
    However, after inspecting a bunch of user support bundles I see that
    we still have occasional connect errors. Rather than increasing the
    timeout, we attempt here to reuse the handles, thus avoid the
    problem almost entirely.

    Signed-off-by: Tiago Castro <[email protected]>

---

    refactor(rebuild): rebuild completion is not an error

    When the rebuild has been complete, if we wait for it this fails because
    the channels are not longer available.
    Instead, simply return the rebuild state, since this is what we want anyway.

    Signed-off-by: Tiago Castro <[email protected]>

Co-authored-by: Tiago Castro <[email protected]>

fix: check valid sct and sc combinations for pi error

Signed-off-by: Diwakar Sharma <[email protected]>

fix: use auto-detected sector size for blockdev

This fixes the behaviour where we pass 512 as sector size if the
disk uri doesn't contain blk_size parameter. This causes pool creation
failure if the underlying disk has a different sector size e.g. 4096.
Instead of passing 512, we now pass 0 which lets spdk detect the
device's sector size and use that value.

Signed-off-by: Diwakar Sharma <[email protected]>

This fixes the behaviour where we pass 512 as sector size if the
disk uri doesn't contain blk_size parameter. This causes pool creation
failure if the underlying disk has a different sector size e.g. 4096.
Instead of passing 512, we now pass 0 which lets spdk detect the
device's sector size and use that value.

Signed-off-by: Diwakar Sharma <[email protected]>
@tiagolobocastro
Copy link
Contributor Author

bors try

bors-openebs-mayastor bot pushed a commit that referenced this pull request Nov 26, 2024
@bors-openebs-mayastor
Copy link

try

Build succeeded:

mayastor-bors and others added 2 commits November 26, 2024 18:24
1755: Reuse Rebuild IO handles r=tiagolobocastro a=tiagolobocastro

    fix(rebuild): reuse rebuild IO handles

    Reuses the rebuild IO handles, rather than attempting to allocate
    them per rebuild task.
    The main issue with handle allocation on the fly is that the target
    may have not cleaned up a previous IO qpair connection, and so the
    connect may fail. We started seeing this more on CI because we forgot
    to cherry-pick a commit increasing the retry delay.
    However, after inspecting a bunch of user support bundles I see that
    we still have occasional connect errors. Rather than increasing the
    timeout, we attempt here to reuse the handles, thus avoid the
    problem almost entirely.

    Signed-off-by: Tiago Castro <[email protected]>

---

    refactor(rebuild): rebuild completion is not an error

    When the rebuild has been complete, if we wait for it this fails because
    the channels are not longer available.
    Instead, simply return the rebuild state, since this is what we want anyway.

    Signed-off-by: Tiago Castro <[email protected]>

Co-authored-by: Tiago Castro <[email protected]>
Signed-off-by: Tiago Castro <[email protected]>
Caters for when the device is /dev/nvmeX but X not the same as the controller!

Signed-off-by: Tiago Castro <[email protected]>
@tiagolobocastro
Copy link
Contributor Author

bors merge

@bors-openebs-mayastor
Copy link

Build succeeded:

@bors-openebs-mayastor bors-openebs-mayastor bot merged commit 0d8ac9c into release/2.7 Nov 27, 2024
4 checks passed
@bors-openebs-mayastor bors-openebs-mayastor bot deleted the cherry-pick branch November 27, 2024 12:21
Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment
Labels
None yet
Projects
None yet
Development

Successfully merging this pull request may close these issues.

3 participants