NFS mount over UDP fails on kernel 6.6

Bug #2069643 reported by Matheus Machado Guilhermino
6
This bug affects 1 person
Affects Status Importance Assigned to Milestone
StarlingX
Fix Released
Undecided
Unassigned

Bug Description

Brief Description
-----------------
When installing an AIO-DX system using kernel 6.6, controller-1 install fails when attempting to execute nfs mounts due to "malformed" UDP response that are too short and contain checksum errors.

Severity
--------
Critical

Steps to Reproduce
------------------
Steps to reproduce:

(1)init repo

repo init -u https://opendev.org/starlingx/manifest -b f/kernel-6.6

repo sync --force-sync

(2) build-pkgs; build-image; start AIO-DX install.

Expected Behavior
------------------
Controller-1 installed and configured

Actual Behavior
----------------
Controller-1 is not installed

Reproducibility
---------------
100%

System Configuration
--------------------
Two node system

Last Pass
---------
Never on kernel 6.6

Timestamp/Logs
--------------
[ 202.496856] controller_config[4072]: mount.nfs: Connection timed out
[ 202.505412] controller_config[5006]: ****************************************************
[ 202.515062] controller_config[5006]: ****************************************************
[ 202.525062] controller_config[5006]: Unable to mount /opt/platform
[ 202.533056] controller_config[5006]: ****************************************************
[ 202.543058] controller_config[5006]: ****************************************************
[ 202.553615] controller_config[3517]: Pausing for 5 seconds...

Test Activity
-------------
Developer Testing

Workaround
----------
Modify nfs_mount script to use TCP instead of UDP

description: updated
Revision history for this message
OpenStack Infra (hudson-openstack) wrote : Fix merged to kernel (f/kernel-6.6)

Reviewed: https://review.opendev.org/c/starlingx/kernel/+/922145
Committed: https://opendev.org/starlingx/kernel/commit/3c6402b23c7b87a6d43a2c0a62fe1720aa620011
Submitter: "Zuul (22348)"
Branch: f/kernel-6.6

commit 3c6402b23c7b87a6d43a2c0a62fe1720aa620011
Author: Matheus Guilhermino <email address hidden>
Date: Mon Jun 17 16:32:12 2024 -0300

    SUNRPC: avoid zero-length UDP packets

    Cherry-picks commit 1d9cabe2817e ("SUNRPC: use request size to
    initialize bio_vec in svc_udp_sendto()") from linux-yocto in order
    to use the proper size when setting up the bio_vec, as otherwise
    only zero-length UDP packets will be sent.

    * commit d9cabe2817e ("SUNRPC: use request size to initialize
      bio_vec in svc_udp_sendto()")
      https://git.yoctoproject.org/linux-yocto/commit/?id=1d9cabe2817e

    Test Plan:
    PASS: Succesful NFS mount during C-1 install of an AIO-DX

    Closes-bug: 2069643

    Change-Id: Iac09d24ff5f3cc4c64690e1c42356d7f57f95d66
    Signed-off-by: Matheus Guilhermino <email address hidden>

Changed in starlingx:
status: New → In Progress
Revision history for this message
OpenStack Infra (hudson-openstack) wrote : Fix merged to kernel (master)

Reviewed: https://review.opendev.org/c/starlingx/kernel/+/923005
Committed: https://opendev.org/starlingx/kernel/commit/c75144224057b36a88700ba11cf65327ca8f48ff
Submitter: "Zuul (22348)"
Branch: master

commit c75144224057b36a88700ba11cf65327ca8f48ff
Author: Matheus Guilhermino <email address hidden>
Date: Mon Jun 17 16:32:12 2024 -0300

    SUNRPC: avoid zero-length UDP packets

    Cherry-picks commit 1d9cabe2817e ("SUNRPC: use request size to
    initialize bio_vec in svc_udp_sendto()") from linux-yocto in order
    to use the proper size when setting up the bio_vec, as otherwise
    only zero-length UDP packets will be sent.

    * commit d9cabe2817e ("SUNRPC: use request size to initialize
      bio_vec in svc_udp_sendto()")
      https://git.yoctoproject.org/linux-yocto/commit/?id=1d9cabe2817e

    Test Plan:
    PASS: Succesful NFS mount during C-1 install of an AIO-DX

    Closes-bug: 2069643

    Change-Id: Iac09d24ff5f3cc4c64690e1c42356d7f57f95d66
    Signed-off-by: Matheus Guilhermino <email address hidden>
    (cherry picked from commit 3c6402b23c7b87a6d43a2c0a62fe1720aa620011)
    Signed-off-by: Jiping Ma <email address hidden>

Changed in starlingx:
status: In Progress → Fix Released
To post a comment you must log in.
This report contains Public information  
Everyone can see this information.

Other bug subscribers

Remote bug watches

Bug watches keep track of this bug in other bug trackers.