NSX plugin: punctual state synchronization might cause timeouts

[Salvatore Orlando]

When performing punctual state synchronization (*) the NSX plugin might trigger the infamous eventlet-mysql deadlock since a NSX API call is performed from within a DB transaction.

This should be avoided, especially since the transaction is not needed in this case.

(*) This can be enabled either via a configuration variable or by explicitly selecting the 'status' field in Neutron APIs.

[OpenStack Infra (hudson-openstack)]

Fix proposed to branch: master
Review: https://review.openstack.org/73033

[OpenStack Infra (hudson-openstack)]

Related fix proposed to branch: master
Review: https://review.openstack.org/73695

[OpenStack Infra (hudson-openstack)]

Reviewed: https://review.openstack.org/73695
Committed: https://git.openstack.org/cgit/openstack/neutron/commit/?id=85ddbde058d8bda0b938eb7a45ef73519a831b3b
Submitter: Jenkins
Branch: master

commit 85ddbde058d8bda0b938eb7a45ef73519a831b3b
Author: armando-migliaccio <email address hidden>
Date: Fri Feb 14 10:59:17 2014 -0800

    Fix request timeout errors during calls to NSX controller

    Sometimes two correlated exception traces are observed in
    the server log for the Neutron Server backed by NSX:
    RequestTimeout (The nsx request has timed out) and
    OperationalError (Lock wait timeout exceeded). This is
    generally described by Guru Salvatore Orlando as the,
    and I quote, the "infamous eventlet-mysql deadlock".

    This patch tries to address the issue by adding a
    cooperative yield in the nsx client code (it’s a good idea
    to call sleep(0) occasionally in any case) and also by
    avoiding the unnecessary spawning of another Greenthread
    within a call that is already executed in Greenthred
    itself.

    Closes-bug: #1267101
    Related-bug: #1279497

    Change-Id: I8e298468fb730f11a66fbd4211121ee7d3e2a548

[OpenStack Infra (hudson-openstack)]

Related fix proposed to branch: stable/havana
Review: https://review.openstack.org/75484

[OpenStack Infra (hudson-openstack)]

Reviewed: https://review.openstack.org/75484
Committed: https://git.openstack.org/cgit/openstack/neutron/commit/?id=c1942623a5ac0be5df83d9511b56bb6009b64e69
Submitter: Jenkins
Branch: stable/havana

commit c1942623a5ac0be5df83d9511b56bb6009b64e69
Author: armando-migliaccio <email address hidden>
Date: Fri Feb 14 10:59:17 2014 -0800

    Fix request timeout errors during calls to NSX controller

    Sometimes two correlated exception traces are observed in
    the server log for the Neutron Server backed by NSX:
    RequestTimeout (The nsx request has timed out) and
    OperationalError (Lock wait timeout exceeded). This is
    generally described by Guru Salvatore Orlando as the,
    and I quote, the "infamous eventlet-mysql deadlock".

    This patch tries to address the issue by adding a
    cooperative yield in the nsx client code (it’s a good idea
    to call sleep(0) occasionally in any case) and also by
    avoiding the unnecessary spawning of another Greenthread
    within a call that is already executed in Greenthred
    itself.

    Closes-bug: #1267101
    Related-bug: #1279497

    Change-Id: I8e298468fb730f11a66fbd4211121ee7d3e2a548
    (cherry picked from commit 85ddbde058d8bda0b938eb7a45ef73519a831b3b)