You are viewing a plain text version of this content. The canonical link for it is here.
Posted to users@jackrabbit.apache.org by Ian Boston <ie...@tfd.co.uk> on 2009/08/27 11:48:15 UTC
InternalError in UserAccessControlProvider
Hi,
JR 1.5.6
I am getting an "Internal Error" from the UserAccessControlProvider
traced with a debugger to line
UserAccessControlProvider .java:506 where the code says
} catch (RepositoryException e) {
// should never get here
log.error("Internal error ", e.getMessage());
}
On inspection the error is
javax.jcr.InvalidItemStateException: c86bf464-509e-4e8a-808a-
fd729cef0be9: the item does not exist anymore
Users are being created and deleted from the security workspace when
this is happening.
Although the comment says it should not happen, it does, regularly.
Is it anything to be concerned about ?
Thanks
Ian
Re: InternalError in UserAccessControlProvider
Posted by Ian Boston <ie...@tfd.co.uk>.
On 27 Aug 2009, at 14:03, Tobias Bocanegra wrote:
> it would be better if you create a junit test case for jackrabbit -
> because we don't use sling for testing.
> regards, toby
Ok, I will have a go at isolating the issue.
Ian
Re: InternalError in UserAccessControlProvider
Posted by Tobias Bocanegra <tr...@day.com>.
On Thu, Aug 27, 2009 at 12:15 PM, Ian Boston<ie...@tfd.co.uk> wrote:
>
> On 27 Aug 2009, at 11:08, Stefan Guggisberg wrote:
>
>>> Is it anything to be concerned about ?
>>
>> without knowing the code in question, i think, yes. the code obviously
>> assumes
>> that this situation should never happen. however, if it does, i guess
>> it's a bug.
>
> This is Sling with some modifications/enhancements/additions.
> Although I have made modifications to other areas of the non-securirty
> access control structure, I dont think I have made any to the
> UserManagerImpl or the UserAccesControlProvider which are direct from the
> 1.5.6 jackrabbit jar.
>
> The test that reproduces this is a Ruby script running outside the JVM,
> single threaded, exercising the Sling http user manager endpoints.
>
> If its a real problem (from your response, I guess it is) I will try and
> reproduce against an unmodified Sling with a simple bash/curl script.
it would be better if you create a junit test case for jackrabbit -
because we don't use sling for testing.
regards, toby
Re: InternalError in UserAccessControlProvider
Posted by Ian Boston <ie...@tfd.co.uk>.
On 27 Aug 2009, at 11:08, Stefan Guggisberg wrote:
>> Is it anything to be concerned about ?
>
> without knowing the code in question, i think, yes. the code
> obviously assumes
> that this situation should never happen. however, if it does, i guess
> it's a bug.
This is Sling with some modifications/enhancements/additions.
Although I have made modifications to other areas of the non-securirty
access control structure, I dont think I have made any to the
UserManagerImpl or the UserAccesControlProvider which are direct from
the 1.5.6 jackrabbit jar.
The test that reproduces this is a Ruby script running outside the
JVM, single threaded, exercising the Sling http user manager endpoints.
If its a real problem (from your response, I guess it is) I will try
and reproduce against an unmodified Sling with a simple bash/curl
script.
Thanks,
Ian
Re: InternalError in UserAccessControlProvider
Posted by Stefan Guggisberg <st...@gmail.com>.
hi ian
On Thu, Aug 27, 2009 at 11:48 AM, Ian Boston<ie...@tfd.co.uk> wrote:
> Hi,
> JR 1.5.6
>
> I am getting an "Internal Error" from the UserAccessControlProvider traced
> with a debugger to line
> UserAccessControlProvider .java:506 where the code says
>
> } catch (RepositoryException e) {
> // should never get here
> log.error("Internal error ", e.getMessage());
> }
>
> On inspection the error is
> javax.jcr.InvalidItemStateException: c86bf464-509e-4e8a-808a-fd729cef0be9:
> the item does not exist anymore
>
> Users are being created and deleted from the security workspace when this is
> happening.
>
> Although the comment says it should not happen, it does, regularly.
>
> Is it anything to be concerned about ?
without knowing the code in question, i think, yes. the code obviously assumes
that this situation should never happen. however, if it does, i guess
it's a bug.
cheers
stefan
>
> Thanks
> Ian
>