|
From: Leif M. <lei...@ta...> - 2011-05-09 09:39:18
|
Peter, The "wrapper -s conf" command is used by the Service Manager to launch the Wrapper as a service. That should never have worked from the command line. You would see something like this: --- Attempting to start Test Wrapper Sample Application as an NT service. Calling StartServiceCtrlDispatcher...please wait. StartServiceControlDispatcher failed! The -s and --service commands should only be called by the Windows ServiceManager to control the Wrapper as a service, and is not designed to be run manually by the user. For help, type bin\wrapper -? --- "wrapper -t conf" is what you want to start the service. I am not sure what could be causing the failed to start problem. That message is usually shown by the Windows Service Manager if the service takes too long to start without reporting its status, but you said that it happens almost immediately. Could you set the wrapper.debug=true property and then send a wrapper.log file that shows this happening? That might give me a bit more information to be able to help further. Cheers, Leif On Mon, May 9, 2011 at 5:49 PM, Peter Beecken <pbe...@am...> wrote: > Dear all, > > Some wrapper services are not starting on our Windows cluster. > They do not even start up as a local service. > The log file tells us "StartServiceControlDispatcher failed!" > > We have a Windows cluster with 5 physical nodes for monitoring purposes. > There are 13 resource groups. > 8 of them run home made Java services using Wrapper 3.2.3. > These resource groups also have a Tivoli backup client. > Originally all 8 monitoring groups ran on servers 3 and 4. > For better load balancing we decided to also use Server1/2/5 for monitoring. > > I replicated all service related registry entries by script, updated the > cluster config to include the new nodes and moved some groups from the > old servers (3 and 4). > > After replicating the service definitions to Server1,2 and 5 the situation > on resource looks like: > > Service/Node SERVER1 SERVER2 SERVER3 SERVER4 SERVER4 > SCHECK_DZA_PRD OK OK OK OK FAILED > SCHECK_MONOFMOG OK FAILED OK OK FAILED > SCHECK_OPODO OK OK OK OK OK > SCHECK_ORA OK OK OK OK FAILED > SCHECK_UNIX5 FAILED FAILED OK OK FAILED > SCHECK_UNIX6 FAILED OK OK OK FAILED > SCHECK_VCS OK OK OK OK FAILED > > FAILED meaning that the the service did not start. > ERROR | wrapper | 2011/05/09 08:14:09 | Unable to start the > SCHECK_DZA_PRD service - > The service did not respond to the start > or control request in a timely fashion. (0x41d) > > Message shows up immediately, few milliseconds delay only. > > Wrapper services are started as: > > G:\SCheck_DZA_PRD\wrapper.exe -s G:\SCheck_DZA_PRD\\conf\wrapper.conf > > For some reason -s was used in the past and worked. > Changed it to -t but results did not change. > > As all resources of a service are in one folder, I copied the folder to > local disk, changed the service command line (regedit and wrapper.conf) > to reflect the correct folder and could start the service. > > Copying the folder to another SAN drive resulted in a working service. > > Copying/renaming the folder on the same SAN drive ended up with the > above error message. > > Configuration: > Server OS System CPUs VCPUs Processor Mem(GB) > SERVER1 2003EE SP2 ProLiant DL380 G5 2 8 x86 Family 6 > 5.2.3790 Model 15 Stepping 11 16 > GenuineIntel ~2666 Mhz > SERVER2 2003EE SP2 ProLiant DL380 G5 2 8 x86 Family 6 > 5.2.3790 Model 15 Stepping 11 > GenuineIntel ~2666 Mhz 16 > SERVER3 2003EE SP2 ProLiant DL580 G5 4 16 x86 Family 6 > 5.2.3790 Model 15 Stepping 11 > GenuineIntel ~2933 Mhz" 16 > SERVER4 2003EE SP2 ProLiant DL580 G5 4 16 x86 Family 6 > 5.2.3790 Model 15 Stepping 11 > GenuineIntel ~2933 Mhz" 16 > SERVER5 2003EE SP2 ProLiant DL580 G5 4 16 x86 Family 6 > 5.2.3790 Model 29 Stepping 1 > GenuineIntel ~2400 Mhz" 8 > > All servers use Veritas and SAN devices (EMC SYMMETRIX) for drives G, H, ... > Local drives are C: D: E: (used for Windows and S/W not related to Wrapper). > Wrapper Version 3.2.3 being used for all Wrapper services. > Also tried 3.8.7 with exactly same results. > > Any idea what I might try next? |