Created attachment 108655 [details] in-memory cluster.conf
Description of problem: ccsd segfaulted while running ccs_test. from the logs... Dec 15 15:18:22 trin-01 ccsd[8482]: Error: unable to evaluate xpath query "/cluster/" Dec 15 15:18:22 trin-01 ccsd[8482]: Error while processing get: Invalid argument Dec 15 15:18:33 trin-01 ccsd[8482]: Stopping ccsd, SIGSEGV received. Version-Release number of selected component (if applicable): kernel-2.6.9-1.906_EL cman-kernel-2.6.9-3.3 ccs-0.9-0 cman-1.0-0.pre5.0 fence-1.3-1 How reproducible: haven't tried... I've only seen it the one time so far Steps to Reproduce: [root@trin-01 fenced]# ccs_test connect Connect successful. Connection descriptor = 3 [root@trin-01 fenced]# ccs_test get 3 /cluster Get successful. Value = < > [root@trin-01 fenced]# ccs_test get 3 /cluster/ ccs_get failed: Invalid argument [root@trin-01 fenced]# ccs_test get 3 /cluster/fence_daemon ccs_get failed: Invalid exchange [root@trin-01 fenced]# ccs_test get 3 /cluster/fence_daemon/ <<-- here is where I noticed that ccsd SEGV'ed in the logs -->> Actual results: segfault Expected results: no segfault Additional info: I had the fence_domain running at the time and had been trying to update the cluster.conf a couple of times. I often had to retry updating ccs because I kept forgetting to update the version number of the config file. I don't know if that had anything to do with this though.
I am unable to reproduce this bug. (I started working on it after the update fixes, which may have played a partial roll.) Generally, an 'invalid exchange' error is due to not being able to receive or send a (complete) package. I've looked at all the places where this error code is set and found nothing. Perhaps it has to do with the previous EINVAL error, but I can't find anything wrong with this either... This did uncover another bug, however. One which only occurred if the working path had been set. This has been fixed.