Details
-
Bug
-
Status: Closed
-
Normal
-
Resolution: Fixed
-
None
-
None
-
None
Description
Impact before the fix of HSTTWO-2305: The cluster node affected always returns a 404 after a request.
Impact after the fix of HSTTWO-2305: The cluster node affected returns a 404 page once.
To reproduce:
- Run a 4-node cluster (I had 4 instances of Hippo Go Green and a MySQL database)
- Introduce traffic on all nodes (I used 1 request per 4 seconds on each node, using jmeter)
- In the channel manager continuously (preferably automated):
> Edit the template
> Save & close
> Publish
When this has run for some time (a few minutes seemed to be enough), the error message appeared. When stopping the publications, this will lead to a 404 error on every request thereafter. When making another publish, this is however solved, and the correct site is rendered again.
The failure never happens on the node you are operating on.
The following error is logged:
-----8<----
28.09.2012 15:26:20 ERROR [org.hippoecm.hst.configuration.hosting.VirtualHostService.<init>():244] Skipping incorrect mount or port mount for mount node '/hst:hst/hst:hosts/dev-localhost/localhost/hst:root'
org.hippoecm.hst.service.ServiceException: There is no preview configuration available for '/hst:hst/hst:configurations/hippogogreen'. Also it is not ending with the convention -preview. Cannot load a configuration for it
at org.hippoecm.hst.configuration.site.HstSiteService.findConfigurationPath(HstSiteService.java:138)
at org.hippoecm.hst.configuration.site.HstSiteService.<init>(HstSiteService.java:72)
at org.hippoecm.hst.configuration.hosting.MountService.<init>(MountService.java:505)
at org.hippoecm.hst.configuration.hosting.VirtualHostService.<init>(VirtualHostService.java:240)
at org.hippoecm.hst.configuration.hosting.VirtualHostsService.<init>(VirtualHostsService.java:207)
at org.hippoecm.hst.configuration.model.HstManagerImpl.buildSites(HstManagerImpl.java:584)
at org.hippoecm.hst.configuration.model.HstManagerImpl.getVirtualHosts(HstManagerImpl.java:249)
at org.hippoecm.hst.container.HstFilter.doFilter(HstFilter.java:266)
at org.apache.catalina.core.ApplicationFilterChain.internalDoFilter(ApplicationFilterChain.java:235)
-----8<----
See HSTTWO-2305 for the implementation of the workaround
See HSTTWO-2311 for some code check-ins regarding this issue and many findings already.
Attachments
Issue Links
- relates to
-
HSTTWO-2311 Fine-grained reloading may fail on cluster nodes when publishing HST changes : part (1)
- Closed
-
HSTTWO-2328 Change the JCRValueProviderImpl to throw a RuntimeRepositoryException in case a unrecoverable exception takes place
- Closed