|
From: John M. <joh...@jb...> - 2005-01-24 15:13:38
|
[sorry for dup - originally sent this to Help forum; just now found out that is old and that I should be sending to this mailing list] I am seeing some weird behavior dealing with the anchorfile. I have my anchorfile set. And under certain conditions, I have seen it work. For example, if I start in -c console mode, it works (anchorfile is created and the JVM will die when I specifically delete it).=20 =20 I can see this work when I install the wrapper as an NT service BUT ONLY if i start the service using the Windows GUI utility (the Services GUI from Administration tools). If, however, I start the Windows service via the -t wrapper option, the JVM starts up fine, but immediately the anchorfile gets deleted and the wrapper immediately shuts the JVM down. It is very weird- I have no idea why or who is deleting that anchor file - but as soon as it gets created (no more than a second or two delay) that anchor file gets deleted.=20 =20 And this only happens if i start the service using the -t. Again, if I start the service using the Winows Services GUI, it all works fine.=20 =20 See below for my debug log - note the time stamps and see that the anchorfile gets deleted rather quickly after startup.=20 =20 Has anyone seen this behavior, or more importantly, how to fix it?=20 =20 ----=20 INFO | jvm 1 | 2005/01/22 00:53:38 | 00:53:38,567 INFO [NamingService] Started jnpPort=3D1099, rmiPort=3D1098, backlog=3D50, = bindAddress=3D/0.0.0.0, Client SocketFactory=3Dnull, Server SocketFactory=3Dorg.jboss.net.sockets.DefaultSocketFactory@ad093076=20 STATUS | wrapper | 2005/01/22 00:53:38 | JBoss [abc-dpl] started.=20 INFO | jvm 1 | 2005/01/22 00:53:40 | 00:53:40,120 INFO [Server] JBoss (MX MicroKernel) [4.0.0 (build: CVSTag=3DJBoss_4_0_0 = date=3D200412201046)] Started in 2s:604ms=20 STATUS | wrapper | 2005/01/22 00:53:41 | Anchor file deleted. Shutting down.=20 DEBUG | wrapper | 2005/01/22 00:53:41 | wrapperStopProcess(0) called.=20 DEBUG | wrapper | 2005/01/22 00:53:41 | Sending stop signal to JVM FWIW: I set wrapper.anchor.poll_interval to something higher (15 seconds) and it still behaves as I describe in my first post. If I start with -t, the JVM stays up for 15 seconds, but the anchor file is gone almost immediately. So, after the 15 seconds are up, since the file is gone, the JVM is shutdown. It looks like as soon as the wrapper spits out the log "wrapper | <my service name> started.", that's when the anchor file gets deleted.=20 =20 Don't know if any of this helps. I looked at the C code but nothing jumped out at me as to a possible problem. Should I write this up as a bug or am I missing something obvious? |