• 1 reply
  • Latest Post - ‏2012-11-28T20:57:36Z by SystemAdmin
1485 Posts

Pinned topic [Beta8.6] stopping a Catalog Server fails intermittently

‏2012-11-17T20:15:10Z |

I started a Catalog Server cluster with 2 members. I stopped one member and I immediately stopped the second member.

Sometimes the second member stop fails with an error:

[11/17/12 21:17:15:396 CET] 00000001 LocationServi I   CWOBJ0075I: The transport type is eXtremeIO.$Transient [originating=;exid=0]:$Transient:Routed to the wron g PlacementService 

for method deregisterObjectGridServers, retry the request. at sun.reflect.NativeConstructorAccessorImpl.newInstance0(Native Method) at sun.reflect.NativeConstructorAccessorImpl.newInstance( at sun.reflect.DelegatingConstructorAccessorImpl.newInstance( at java.lang.reflect.Constructor.newInstance( at at at at at at at sun.reflect.NativeMethodAccessorImpl.invoke0(Native Method) at sun.reflect.NativeMethodAccessorImpl.invoke( at sun.reflect.DelegatingMethodAccessorImpl.invoke( at java.lang.reflect.Method.invoke( at at $Proxy0.tearDownServers(Unknown Source) at at at [11/17/12 21:17:15:474 CET] 00000001 StopServerCli E   CWOBJ2413E: The attempt to stop the server was unsuccessful.

When I tried again to stop the second member, it stopped successfully.
Is this an expected behaviour ? Is there a way to prevent this ?
Updated on 2012-11-28T20:57:36Z at 2012-11-28T20:57:36Z by SystemAdmin
  • SystemAdmin
    1485 Posts

    Re: [Beta8.6] stopping a Catalog Server fails intermittently


    Yes, I would expect that this may happen. When you have multiple catalog servers, one is elected as primary amongst the peers. The non-primary catalog servers communicate with the primary server for several One of N services, the Placement Service being one. When you stop the second, non-primary catalog server it tries to communicate with the primary server (who at this point is going down/already down and a new primary hasn't been elected). It's this communication from the non-primary to the primary that is failing.

    The reason it works later is that the remaining catalog server has been elected the primary and so when stopping can find the primary (itself) to get the placement service.

    Specifying both servers on a single invocation, as opposed to using two different invocations should stop both servers.

    Let me know if you have further questions.