setting reclaim_age longer than default doesn't work

[clayg]

You can set it shorter and it *mostly* works - but if you try to set it *longer* than one week you'll find that every time it makes a REPLICATE call to an object-server that triggers a suffix rehash it'll call clean_ondisk_files with that manager reclaim_age which will be the default ONE_WEEK.

The work around is of course to set reclaim_age in the [DEFAULT] section since the object-server already passes it's conf through to the DiskFileManager.

This should be the documented deployment configuration.

[Mahati Chamarthy (mahati-chamarthy)]

Associated fix -> https://review.openstack.org/#/c/374419/

[OpenStack Infra (hudson-openstack)]

Reviewed: https://review.openstack.org/374419
Committed: https://git.openstack.org/cgit/openstack/swift/commit/?id=69f7be99a6e090a251412a6925e0f29946818c6a
Submitter: Jenkins
Branch: master

commit 69f7be99a6e090a251412a6925e0f29946818c6a
Author: Mahati Chamarthy <email address hidden>
Date: Mon Jul 25 20:10:44 2016 +0530

    Move documented reclaim_age option to correct location

    The reclaim_age is a DiskFile option, it doesn't make sense for two
    different object services or nodes to use different values.

    I also driveby cleanup the reclaim_age plumbing from get_hashes to
    cleanup_ondisk_files since it's a method on the Manager and has access
    to the configured reclaim_age. This fixes a bug where finalize_put
    wouldn't use the [DEFAULT]/object-server configured reclaim_age - which
    is normally benign but leads to weird behavior on DELETE requests with
    really small reclaim_age.

    There's a couple of places in the replicator and reconstructor that
    reach into their manager to borrow the reclaim_age when emptying out
    the aborted PUTs that failed to cleanup their files in tmp - but that
    timeout doesn't really need to be coupled with reclaim_age and that
    method could have just as reasonably been implemented on the Manager.

    UpgradeImpact: Previously the reclaim_age was documented to be
    configurable in various object-* services config sections, but that did
    not work correctly unless you also configured the option for the
    object-server because of REPLICATE request rehash cleanup. All object
    services must use the same reclaim_age. If you require a non-default
    reclaim age it should be set in the [DEFAULT] section. If there are
    different non-default values, the greater should be used for all object
    services and configured only in the [DEFAULT] section.

    If you specify a reclaim_age value in any object related config you
    should move it to *only* the [DEFAULT] section before you upgrade. If
    you configure a reclaim_age less that your consistency window you are
    likely to be eaten by a Grue.

    Closes-Bug: #1626296

    Change-Id: I2b9189941ac29f6e3be69f76ff1c416315270916
    Co-Authored-By: Clay Gerrard <email address hidden>

[OpenStack Infra (hudson-openstack)]

This issue was fixed in the openstack/swift 2.13.0 release.