Description of problem: ----------------------- When RDMA init fails, the resources are not cleaned up properly. Consequence is that glusterd crashes, when taking glusterd statedump. Version-Release number of selected component (if applicable): ------------------------------------------------------------- glusterfs-3.6.0.45-1.el6rhs How reproducible: ----------------- Consistent Steps to Reproduce: ------------------- 1. Start glusterd on non-RDMA setup Actual results: --------------- RDMA init fails to free the resources Expected results: ------------------ When RDMA init failed, all the associated resources should be cleaned up properly
This issue blocks getting glusterd statedump and hence blocks verification of that bug 1181044
https://code.engineering.redhat.com/gerrit/#/c/42282/
Root cause identified as: When an rdma init is failed, the created private variable is not freeing the resources, that's making the the address out of bound.
Earlier without the fix, getting the glusterd statedump would crash, as the variable reached out-of-bounds With this fix, glusterd statedump doesn't crash and hence verifies that there is no out-of-bounds error hit, when RDMA is not initialized. Tested with glusterfs-3.6.0.46-1.el6rhs
Since the problem described in this bug report should be resolved in a recent advisory, it has been closed with a resolution of ERRATA. For information on the advisory, and where to find the updated files, follow the link below. If the solution does not work for you, open a new bug report. https://rhn.redhat.com/errata/RHBA-2015-0682.html