[NETVIRT-1291] New transaction leak in netvirt.fibmanager.VrfEntryListener Created: 01/Jun/18  Updated: 25/Jul/18  Resolved: 23/Jul/18

Status: Resolved
Project: netvirt
Component/s: None
Affects Version/s: Oxygen
Fix Version/s: Oxygen-SR2, Fluorine

Type: Bug Priority: Medium
Reporter: Sai Sindhur Malleni Assignee: Stephen Kitt
Resolution: Done Votes: 0
Labels: None
Remaining Estimate: Not Specified
Time Spent: Not Specified
Original Estimate: Not Specified

Issue Links:
Blocks
blocks NETVIRT-1384 Umbrella: Numerous new transaction leaks Resolved

 Description   

Running performance and scale tests that create 100s of neutron resources like networks, subnets, ports and routers as well as launching VMs on the networks and waiting for the environment to quiesce, we observe some new transaction leaks based on Michael Vorburger's mdsal-trace tool.

24x transactions opened here, which are not closed:
  (...)
  org.opendaylight.controller.md.sal.binding.impl.BindingDOMDataBrokerAdapter.newWriteOnlyTransaction(BindingDOMDataBrokerAdapter.java:69)
  (...)
  org.opendaylight.netvirt.fibmanager.VrfEntryListener.lambda$populateFibOnNewDpn$17(VrfEntryListener.java:1575)
  org.opendaylight.infrautils.jobcoordinator.internal.JobCoordinatorImpl$MainTask.runWithUncheckedExceptionLogging(JobCoordinatorImpl.java:416)
  org.opendaylight.infrautils.utils.concurrent.LoggingUncaughtThreadDeathContextRunnable.run(LoggingUncaughtThreadDeathContextRunnable.java:60)

 Comments   
Comment by Michael Vorburger [ 01/Jun/18 ]

That is already plugged on master by https://git.opendaylight.org/gerrit/#/c/71591/. The stable/oxygen branch of VrfEntryListener for some reason seems to be quite different from master, and still has some dataBroker.newWriteOnlyTransaction() - or may be recently re-introduced by E// friends.

skitt shall I cherry-pick your c/71591 from master to stable/oxygen or do you want to?

shague FYI.

Comment by Stephen Kitt [ 01/Jun/18 ]

I’ll take care of the cherry-pick.

Comment by Michael Vorburger [ 23/Jul/18 ]

closing this one, as both oxygen and master changes are merged now.

Comment by Michael Vorburger [ 25/Jul/18 ]

NETVIRT-1384 also has this one (among many others). That is surprising, given this has fix version Oxygen-SR2, isn't it? Perhaps the the fix version is just wrong - I'll change this to SR3 (and Fluorine), and cross fingers.

Comment by Michael Vorburger [ 25/Jul/18 ]

Actually what shows up in NETVIRT-1384 also is also in VrfEntryListener, but not identical - it's 258x VrfEntryListener$CleanupVpnInterfaceWorker.call(VrfEntryListener.java:1295), which uses callWithNewReadWriteTransactionAndSubmit .. already managed. So I'm changing the Fix Version back to SR2 which was probably right. All good then, nothing to see here.

Generated at Wed Feb 07 20:23:42 UTC 2024 using Jira 8.20.10#820010-sha1:ace47f9899e9ee25d7157d59aa17ab06aee30d3d.