[ovirt-devel] [ OST Failure Report ] [ oVirt 4.2 ] [ 2018-04-04 ] [006_migrations.prepare_migration_attachments_ipv6]

Tue Apr 24 11:00:04 UTC 2018

Ravi's patch is in, but a similar problem remains, and the test cannot
be put back into its place.

It seems that while Vdsm was taken down, a couple of getCapsAsync
requests queued up. At one point, the host resumed its connection,
before the requests have been cleared of the queue. After the host is
up, the following tests resume, and at a pseudorandom point in time,
an old getCapsAsync request times out and kills our connection.

I believe that as long as ANY request is on flight, the monitoring
lock should not be released, and the host should not be declared as
up.

On Wed, Apr 11, 2018 at 1:04 AM, Ravi Shankar Nori <rnori at redhat.com> wrote:
> This [1] should fix the multiple release lock issue
>
> [1] https://gerrit.ovirt.org/#/c/90077/
>
> On Tue, Apr 10, 2018 at 3:53 PM, Ravi Shankar Nori <rnori at redhat.com> wrote:
>>
>> Working on a patch will post a fix
>>
>> Thanks
>>
>> Ravi
>>
>> On Tue, Apr 10, 2018 at 9:14 AM, Alona Kaplan <alkaplan at redhat.com> wrote:
>>>
>>> Hi all,
>>>
>>> Looking at the log it seems that the new GetCapabilitiesAsync is
>>> responsible for the mess.
>>>
>>> - 08:29:47 - engine loses connectivity to host
>>> 'lago-basic-suite-4-2-host-0'.
>>>
>>> - Every 3 seconds a getCapabalititiesAsync request is sent to the host
>>> (unsuccessfully).
>>>
>>>      * before each "getCapabilitiesAsync" the monitoring lock is taken
>>> (VdsManager,refreshImpl)
>>>
>>>      * "getCapabilitiesAsync" immediately fails and throws
>>> 'VDSNetworkException: java.net.ConnectException: Connection refused'. The
>>> exception is caught by
>>> 'GetCapabilitiesAsyncVDSCommand.executeVdsBrokerCommand' which calls
>>> 'onFailure' of the callback and re-throws the exception.
>>>
>>>          catch (Throwable t) {
>>>             getParameters().getCallback().onFailure(t);
>>>             throw t;
>>>          }
>>>
>>>     * The 'onFailure' of the callback releases the "monitoringLock"
>>> ('postProcessRefresh()->afterRefreshTreatment()-> if (!succeeded)
>>> lockManager.releaseLock(monitoringLock);')
>>>
>>>     * 'VdsManager,refreshImpl' catches the network exception, marks
>>> 'releaseLock = true' and tries to release the already released lock.
>>>
>>>       The following warning is printed to the log -
>>>
>>>       WARN  [org.ovirt.engine.core.bll.lock.InMemoryLockManager]
>>> (EE-ManagedThreadFactory-engineScheduled-Thread-53) [] Trying to release
>>> exclusive lock which does not exist, lock key:
>>> 'ecf53d69-eb68-4b11-8df2-c4aa4e19bd93VDS_INIT'
>>>
>>>
>>> - 08:30:51 a successful getCapabilitiesAsync is sent.
>>>
>>> - 08:32:55 - The failing test starts (Setup Networks for setting ipv6).
>>>
>>>
>>>     * SetupNetworks takes the monitoring lock.
>>>
>>> - 08:33:00 - ResponseTracker cleans the getCapabilitiesAsync requests
>>> from 4 minutes ago from its queue and prints a VDSNetworkException: Vds
>>> timeout occured.
>>>
>>>       * When the first request is removed from the queue
>>> ('ResponseTracker.remove()'), the 'Callback.onFailure' is invoked (for the
>>> second time) -> monitoring lock is released (the lock taken by the
>>> SetupNetworks!).
>>>
>>>       * The other requests removed from the queue also try to release the
>>> monitoring lock, but there is nothing to release.
>>>
>>>       * The following warning log is printed -
>>>         WARN  [org.ovirt.engine.core.bll.lock.InMemoryLockManager]
>>> (EE-ManagedThreadFactory-engineScheduled-Thread-14) [] Trying to release
>>> exclusive lock which does not exist, lock key:
>>> 'ecf53d69-eb68-4b11-8df2-c4aa4e19bd93VDS_INIT'
>>>
>>> - 08:33:00 - SetupNetwork fails on Timeout ~4 seconds after is started.
>>> Why? I'm not 100% sure but I guess the late processing of the
>>> 'getCapabilitiesAsync' that causes losing of the monitoring lock and the
>>> late + mupltiple processing of failure is root cause.
>>>
>>>
>>> Ravi, 'getCapabilitiesAsync' failure is treated twice and the lock is
>>> trying to be released three times. Please share your opinion regarding how
>>> it should be fixed.
>>>
>>>
>>> Thanks,
>>>
>>> Alona.
>>>
>>>
>>>
>>>
>>>
>>>
>>> On Sun, Apr 8, 2018 at 1:21 PM, Dan Kenigsberg <danken at redhat.com> wrote:
>>>>
>>>> On Sun, Apr 8, 2018 at 9:21 AM, Edward Haas <ehaas at redhat.com> wrote:
>>>>>
>>>>>
>>>>>
>>>>> On Sun, Apr 8, 2018 at 9:15 AM, Eyal Edri <eedri at redhat.com> wrote:
>>>>>>
>>>>>> Was already done by Yaniv - https://gerrit.ovirt.org/#/c/89851.
>>>>>> Is it still failing?
>>>>>>
>>>>>> On Sun, Apr 8, 2018 at 8:59 AM, Barak Korren <bkorren at redhat.com>
>>>>>> wrote:
>>>>>>>
>>>>>>> On 7 April 2018 at 00:30, Dan Kenigsberg <danken at redhat.com> wrote:
>>>>>>> > No, I am afraid that we have not managed to understand why setting
>>>>>>> > and
>>>>>>> > ipv6 address too the host off the grid. We shall continue
>>>>>>> > researching
>>>>>>> > this next week.
>>>>>>> >
>>>>>>> > Edy, https://gerrit.ovirt.org/#/c/88637/ is already 4 weeks old,
>>>>>>> > but
>>>>>>> > could it possibly be related (I really doubt that)?
>>>>>>> >
>>>>>
>>>>>
>>>>> Sorry, but I do not see how this problem is related to VDSM.
>>>>> There is nothing that indicates that there is a VDSM problem.
>>>>>
>>>>> Has the RPC connection between Engine and VDSM failed?
>>>>>
>>>>
>>>> Further up the thread, Piotr noticed that (at least on one failure of
>>>> this test) that the Vdsm host lost connectivity to its storage, and Vdsm
>>>> process was restarted. However, this does not seems to happen in all cases
>>>> where this test fails.
>>>>
>>>> _______________________________________________
>>>> Devel mailing list
>>>> Devel at ovirt.org
>>>> http://lists.ovirt.org/mailman/listinfo/devel
>>>
>>>
>>
>