Hi reto I have done some performance tests (GET requests) with clerezza in the last days. Since Clerezze-320 is resolved the performance of ssp is massively improved (thanks a lot for that). But sometimes when the maximum of concurrent users (which clerezza can handle) is reached rdflists are going to be corrupt - it doesn't matter if there are persistent rdflists (like a site navigation/breadcrumbs) or temporary lists (usermanager.webinterface). Unfortunately I can not reproduce this regularly and after a hard restart the lists seems to be ok again but since clerezza could not shutdown correctly I can not say if this is because tdb could not save the data or if this is just a temporary/cachning problem of rdflists.
I have created an issue (https://issues.apache.org/jira/browse/CLEREZZA-321) but I am not sure how to handle this issue - I don't get it how a GET request (read action) can make an rdflist corrupt? Any idea? Thanks for your help. Tsuy
