Community
Participate
Working Groups
Start New Connecton Wizard / Discovery. Choose Multicast. Pick a Linux host (ssh-sftp). Finish. --> Host looks good, but when trying to expand it, it is empty. --> Probably due to adapter loading issue. -----------Enter bugs above this line----------- TM 3.0M6 testing installation : eclipse-SDK-3.4M6 (I20080330-1350), cdt-4.0.0M5, emf-2.4.0M5 DSF-N20071113, ECF-2.0m5, PHPEclipse-1.2.0.20080208, Releng.Tools-3.4M5, Subversive-0.7.0.v20080214, SWT-MemMonitor, WR-Retriever-3.0.v20070604, Mylyn-2.3.0.I20080214-1500, RXTX-2.1-7r3b, J9Launching RSE install : RSE HEAD 3.0M6 candidate in workspace, TM-terminal, TM-discovery java.runtime : Sun 1.6.0_02-b06 -Xmx512m -XX:MaxPermSize=128m os.name: : Windows XP 5.1, Service Pack 2 ------------------------------------------------ systemtype : Windows-local, Dstore-win, Dstore-linux targetos : Red Hat Enterprise Linux WS release 4 (Nahant Update 3) targetuname : Linux parser 2.6.9-34.EL #1 i686 athlon i386 GNU/Linux targetvm : Sun Java HotSpot(TM) Client VM (build 1.4.2_12-b03, mixed mode) ------------------------------------------------
It looks like a regression issue, possibly in the registry, because after creating the empty discovery systemtype in ServiceDiscoveryWizard.performFinish() line: 111 , the discovered subsystems are correctly identified and attempted to be attached in ServiceDiscoveryWizard.performFinish() line: 176
Dave - could you have a look at this, please? The problem might be due to the lazy filter loading as per bug 197036 or probably bug 218304.
Bulk update of target milestone
If possible, I'd like some investigation of this before the 3.0 release, such that we understand whether this is a regression or what kind of migration needs to be done. Discovery was working OK in 2.0, it has not been changed at all, and it fails now in 3.0 -- I'd like to understand what caused the failure such that we can either fix the regression or document it in migration notes. An actual fix might go into 3.0.1 but I'd like to understand what causes the issue, just to make sure that adopters don't fall into this regression unnecessarily. Do you think you could squeeze such an investigation in?
This still fails with 3.0.1 -- since Discovery is totally unusable with this, I think we should get this fixed. What's more, we still do not understand what change in RSE Core made the unchanged Discovery code fail. Dave, if you don't think you can get this done by M3, please reassign.
I had an initial look at this but got lost. From the SystemRegistry point of view, the subsystems seem to be created properly. Still, when trying to expand the created host, they are missing. DaveM could you have a look at this? I'm quite sure it is *not* an adapter loading issue since it happens consistently after some time working with other connections in RSE, such that the adapters should be available.
(In reply to comment #6) > I had an initial look at this but got lost. From the SystemRegistry point of > view, the subsystems seem to be created properly. Still, when trying to expand > the created host, they are missing. > > DaveM could you have a look at this? I'm quite sure it is *not* an adapter > loading issue since it happens consistently after some time working with other > connections in RSE, such that the adapters should be available. > I run into a couple problems right now when trying this out. First, I notice there is a null pointer exception if I try to finish the discovery new connection wizard without going to the 2nd page. This occurs in ServiceDiscoverWizard here: public boolean performFinish() { IExtensionPoint ep = Platform.getExtensionRegistry().getExtensionPoint("org.eclipse.rse.core","subsystemConfigurations"); //$NON-NLS-1$ //$NON-NLS-2$ IConfigurationElement[] ce = ep.getConfigurationElements(); SystemRefreshAllAction systemRefreshAllAction = new SystemRefreshAllAction(null); ISystemRegistry registry = RSECorePlugin.getTheSystemRegistry(); String[] addresses = serviceDiscoveryPage.getAddresses(); .... } The serviceDiscoveryPage is null. The second limit I have right now is that none of the hosts I try connect with seem to have any services (i.e. the services page is always empty). Are there known hosts on the internet I can try this with or do I need to setup something myself on a server?
For (1) this is a separate bug, nice catch, could you file a bugzilla for it? For (2) the Fedora Core 5 or later and Redhat Enterprise WS 4 linux OS's usually run a zeroconf server so they do show their SSH servers. As an alternative, I think you could install some Apple software on Windows for testing. I'd need to look up where that was.
(In reply to comment #8) > For (1) this is a separate bug, nice catch, could you file a bugzilla for it? > > For (2) the Fedora Core 5 or later and Redhat Enterprise WS 4 linux OS's > usually run a zeroconf server so they do show their SSH servers. As an > alternative, I think you could install some Apple software on Windows for > testing. I'd need to look up where that was. > I've opened bug 248777 for the first issue.
I don't see us investing in discovery any more.
