• Open Menu Close Menu
  • Apple
  • Shopping Bag
  • Apple
  • Mac
  • iPad
  • iPhone
  • Watch
  • TV
  • Music
  • Support
  • Search apple.com
  • Shopping Bag

Lists

Open Menu Close Menu
  • Terms and Conditions
  • Lists hosted on this site
  • Email the Postmaster
  • Tips for posting to public mailing lists
Temporarily deaf instances
[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]

Temporarily deaf instances


  • Subject: Temporarily deaf instances
  • From: Gary Teter <email@hidden>
  • Date: Mon, 24 Mar 2008 11:56:54 -0700

Recently my instances have been exhibiting some really strange behavior. A couple times a day an instance will simply stop responding to requests for a period of time ranging from a minute to ten minutes. During this period, top claims the instance is using 2.4% or so of processor. Other instances on the same server continue running just fine.

If I issue a kill -QUIT to get a thread dump, I get nothing.... Until the instance decides to wake up from whatever dreaming it's been doing, several minutes later. At that moment I get the thread dump, and the instance logs all of the requests that it should have received during that period, with NSLog timestamps at the moment of re-awakening. Monitor records this as a death, but once the instance comes back it goes back to properly handling requests.

At first I suspected a deadlocking issue, but running with concurrent request handling off doesn't make the weirdness go away. Also, the eventual thread dump doesn't show any deadlocks. (And, what kind of deadlock eventually works itself out?) What really makes me curious is that the adaptor apparently doesn't notice that the requests aren't being handled and continues to send them to the instance. Receive timeout is 60 seconds, all other settings are default.

WebObjects 5.3.3, deployed on Intel Xserves running Mac OS X Server 10.4.10 and 10.4.11. We're using an ERXRemoteSynchronizer-derived change notification system over JMS (openjms-0.7.7-beta-1 running on a different box), and a couple of memcached daemons on another box. (We tried various configurations using jgroups via UDP and/or TCP but it seemed to make the problem worse.)

Any insight is greatly appreciated!

--
WireHose: Smart metadata and personalization for WebObjects.
http://wirehose.com/


_______________________________________________ Do not post admin requests to the list. They will be ignored. Webobjects-dev mailing list (email@hidden) Help/Unsubscribe/Update your Subscription: This email sent to email@hidden
  • Follow-Ups:
    • Re: Temporarily deaf instances
      • From: Mike Schrag <email@hidden>
  • Prev by Date: Re: Kerberos auth
  • Next by Date: Re: Kerberos auth
  • Previous by thread: Re: IE ignoring the anchor, Firefox ok
  • Next by thread: Re: Temporarily deaf instances
  • Index(es):
    • Date
    • Thread