search

enonic / xp

Enonic XP
https://enonic.com
GNU General Public License v3.0
201 stars 34 forks source link

Satisfied WebDispatcherServlet #6517

Closed GlennRicaud closed 4 years ago

GlennRicaud commented 6 years ago

Happened when restarting one node of a 3nodes cluster in RED state. The 12 following OSGI services were 'SATISFIED': com.enonic.xp.web.impl.exception.ExceptionMapperImpl com.enonic.xp.web.impl.handler.WebDispatcherServlet com.enonic.xp.admin.impl.portal.AdminPortalHandler com.enonic.xp.portal.impl.app.AppHandler com.enonic.xp.portal.impl.error.ErrorHandlerScriptFactoryImpl com.enonic.xp.portal.impl.exception.ExceptionRendererImpl com.enonic.xp.portal.impl.handler.attachment.AttachmentHandler com.enonic.xp.portal.impl.handler.identity.IdentityHandler com.enonic.xp.portal.impl.handler.mapping.MappingHandler com.enonic.xp.portal.impl.handler.portal.PortalHandler com.enonic.xp.portal.impl.handler.service.ServiceHandler com.enonic.xp.portal.impl.websocket.WebSocketManagerImpl

Also to note. There was an exception on startup for this node:

23:45:17.473 ERROR c.e.x.c.i.app.ApplicationServiceImpl - Cannot install application [no.gjensidige.intra.redirect]
java.lang.NullPointerException: null
    at com.enonic.xp.core.impl.app.ApplicationServiceImpl.storedApplicationIsStarted(ApplicationServiceImpl.java:448)
    at com.enonic.xp.core.impl.app.ApplicationServiceImpl.doInstallStoredApplications(ApplicationServiceImpl.java:212)

The Enonic XP version (<6.15) is to be confirmed. It matters when trying to reproduce since the startup has been changed with the new cluster behaviour in 6.15

GlennRicaud commented 4 years ago

The problem I have noticed is when admin rest services are created while being queried.

TO reproduce: Restart a node and and bombard of requests the admin rest endpoints

rymsha commented 4 years ago

@sgauruseu @jsi @alansemenov have you ben able to reproduce it recently?

Could not find resource for full path http://localhost:8080/admin/rest ...

alansemenov commented 4 years ago

Not for a long while. Can be closed, I think

rymsha commented 4 years ago

Let's wait for @sgauruseu comment

sgauruseu commented 4 years ago

This problem did not reproduce for me when I used clusters with two nodes, I think that this does not reproduce with three nodes also

rymsha commented 4 years ago

Problem not reproduced, does't mean it disappeared :) By looking closer on affected components I could see that they are either non-immediate (so they stay SATISFIED until other components request them), or they depend (sometimes indirectly) on ContentService. So this is definitely same as #8173

the only difference is that before OSGi logs were not appearing (fixed only in 7.2)