sharder fails when drive is unmounted

[John Dickinson]

The container sharder process fails if a drive is unmounted on the current node. This is because the internal `swift.container.sharder._check_node()` method is not properly handling the response from `check_drive()`. `check_drive()` returns a string if successful and raises a ValueError otherwise. `_check_node()` is simply checking if the response is Falsy and not handling the ValueError at all.

[OpenStack Infra (hudson-openstack)]

Reviewed: https://review.openstack.org/621727
Committed: https://git.openstack.org/cgit/openstack/swift/commit/?id=c26d67efcf44c813034f59b1df78d59fdb4a50e7
Submitter: Zuul
Branch: master

commit c26d67efcf44c813034f59b1df78d59fdb4a50e7
Author: John Dickinson <email address hidden>
Date: Mon Dec 3 14:22:59 2018 -0800

    fixed _check_node() in the container sharder

    Previously, _check_node() wouldn't catch the raise ValueError when
    a drive was unmounted. Therefore the error would bubble up, uncaught,
    and stop the shard cycle. The practical effect is that an unmounted
    drive on a node would prevent sharding for happening.

    This patch updates _check_node() to properly use the check_drive()
    method. Furthermore, the _check_node() return value has been modified
    to be more similar to what check_drive() actually returns. This
    should help prevent similar errors from being introduced in the future.

    Closes-Bug: #1806500

    Change-Id: I3da9b5b120a5980e77ef5c4dc8fa1697e462ce0d

[OpenStack Infra (hudson-openstack)]

Fix proposed to branch: stable/rocky
Review: https://review.openstack.org/624498

[OpenStack Infra (hudson-openstack)]

This issue was fixed in the openstack/swift 2.20.0 release.

[OpenStack Infra (hudson-openstack)]

Reviewed: https://review.openstack.org/624498
Committed: https://git.openstack.org/cgit/openstack/swift/commit/?id=e237328f651e709e907e2fa4b98d9de4bc4f6e20
Submitter: Zuul
Branch: stable/rocky

commit e237328f651e709e907e2fa4b98d9de4bc4f6e20
Author: John Dickinson <email address hidden>
Date: Mon Dec 3 14:22:59 2018 -0800

    fixed _check_node() in the container sharder

    Previously, _check_node() wouldn't catch the raise ValueError when
    a drive was unmounted. Therefore the error would bubble up, uncaught,
    and stop the shard cycle. The practical effect is that an unmounted
    drive on a node would prevent sharding for happening.

    This patch updates _check_node() to properly use the check_drive()
    method. Furthermore, the _check_node() return value has been modified
    to be more similar to what check_drive() actually returns. This
    should help prevent similar errors from being introduced in the future.

    Closes-Bug: #1806500

    Change-Id: I3da9b5b120a5980e77ef5c4dc8fa1697e462ce0d
    (cherry picked from commit c26d67efcf44c813034f59b1df78d59fdb4a50e7)

[OpenStack Infra (hudson-openstack)]

This issue was fixed in the openstack/swift 2.19.1 release.