 |
 |
Hi all, OK, I modified R-GMA test so now it should work reliably both on LCG-2_3_x and LCG-2_4_0. It is now much more simple and easy to understand as it executes only a single check script that goes with R-GMA. Szabolcs: Your site now probably passes R-GMA test. Piotr Hernath Szabolcs wrote: > Hi Piotr, > > thanks for the quick reply. I know it is not critical (yet), but I > just wanted the R-GMA test to show 'green' at least once so badly :-) > Cheers, > > Szabolcs > > On Thu, 14 Apr 2005, Piotr Nyczyk wrote: > >> Dear all, >> >> Unfortunatelly the thing is, that new R-GMA (in LCG-2_4_0) is not fully >> backward compatible. I will have to modify the test in SFT to make it >> work both with old and new version. For the time being the test can't be >> reliable (so that's why it is not a critical test according to SFT >> default configuration). >> >> Piotr >> >> Hernath Szabolcs wrote: >> >>> Dear List, >>> >>> I seem to be out of luck with R-GMA. Todays sft test still failed, >>> and I >>> don't really understand why. Details: >>> >>> 1. Pros >>> >>> The MON box seems to be running the servlets fine: >>> [root@grid154 root]# /opt/edg/sbin/test/edg-rgma-check >>> >>> *** Running R-GMA server tests on grid154.kfki.hu *** >>> >>> Checking servlets... >>> Connecting to http://grid154.kfki.hu:8080/R-GMA/ConsumerServlet:OK >>> Connecting to streaming port 8088 on grid154.kfki.hu:OK >>> Connecting to >>> http://grid154.kfki.hu:8080/R-GMA/StreamProducerServlet:OK >>> Connecting to >>> http://grid154.kfki.hu:8080/R-GMA/LatestProducerServlet:OK >>> Connecting to http://grid154.kfki.hu:8080/R-GMA/DBProducerServlet:OK >>> Connecting to >>> http://grid154.kfki.hu:8080/R-GMA/CanonicalProducerServlet:OK >>> Connecting to http://grid154.kfki.hu:8080/R-GMA/ArchiverServlet:OK >>> >>> *** R-GMA server test successful *** >>> >>> The client side looks all right as well: >>> [root@grid121 root]# /opt/edg/sbin/test/edg-rgma-run-examples >>> >>> *** Running R-GMA client tests on grid121.kfki.hu *** >>> >>> Checking C API: Done. >>> Success >>> Checking C++ API: Success >>> Checking Python API: Success >>> Checking Java API: Success >>> >>> Checking for safe arrival of tuples, please wait... Success >>> >>> *** R-GMA client test successful *** >>> >>> 2. Cons >>> >>> The sft r-gma test complains about: >>> ------------------ >>> +/opt/edg/sbin/test/edg-rgma-check >>> ./testJob.sh: line 318: /opt/edg/sbin/test/edg-rgma-check: No such >>> file or >>> directory >>> + result_check=127 >>> ------------------ >>> The file /opt/edg/sbin/test/edg-rgma-check does not exist on the WNs. >>> Neither on the service nodes for that matter, except for the MON. As >>> this >>> checks the server part, I don't see why it should be present anywhere >>> else, either... >>> >>> Another complaint: >>> ---------------- >>> R-GMA client config file: /opt/edg/var/edg-rgma/rgma.props >>> >>> cat: /opt/edg/var/edg-rgma/rgma.props: No such file or directory >>> ---------------- >>> Actually, it is not present on ANY node at all. Is it OK? If not, how >>> should this file be created? >>> >>> The other perts of the test, publishing/retrieving test data worked out >>> fine as well. In my view, our R-GMA situation is functionally OK. What >>> could be the problem then? On a sidenote, the tests described in >>> http://grid-deployment.web.cern.ch/grid-deployment/documentation/ >>> LCG2-Site-Testing/LCG2-Site-Testing.html#SECTION00060000000000000000 >>> >>> (basically the successful server/client check as above) is fine, too. >>> So, the question is, why do the tests fail? >>> Any hints are welcome, >>> >>> Cheers >>> >>> Szabolcs >>> (BUDAPEST admin) >> >>