Bug 1722187

Summary: Glusterd Seg faults (sig 11) when RDMA used with MLNX_OFED
Product: [Community] GlusterFS Reporter: ryan
Component: glusterdAssignee: bugs <bugs>
Status: CLOSED WONTFIX QA Contact:
Severity: high Docs Contact:
Priority: medium    
Version: 6CC: atumball, bugs
Target Milestone: ---   
Target Release: ---   
Hardware: x86_64   
OS: Linux   
Whiteboard:
Fixed In Version: Doc Type: If docs needed, set a value
Doc Text:
Story Points: ---
Clone Of: Environment:
Last Closed: 2019-07-15 07:24:19 UTC Type: Bug
Regression: --- Mount Type: ---
Documentation: --- CRM:
Verified Versions: Category: ---
oVirt Team: --- RHEL 7.3 requirements from Atomic Host:
Cloudforms Team: --- Target Upstream Version:
Embargoed:
Attachments:
Description Flags
Glusterd --debug output none

Description ryan 2019-06-19 15:53:24 UTC
Created attachment 1582317 [details]
Glusterd --debug output

Description of problem:
Glusterd service fails with signal 11 after installing MLNX_OFED packages when RDMA used for transport.

Version-Release number of selected component (if applicable):
Gluster 6.1

How reproducible:
100% on 2/2 nodes

Steps to Reproduce:
1. Install Glusterfs-server & glusterfs-rdma (6.1)
2. Install MLNX_OFED packages with './mlnxofedinstall --all'

Installing with --all flag installs the following packages:
libibverbs, libibumad,
librdmacm, mft, mstflint, diagnostic tools, OpenSM,
ib-bonding, MVAPICH, Open MPI, MPI tests, MPI selector, perftest, sdpnetstat and libsdp srptools, rdstools, static and dynamic libraries

3.Create Gluster volume with RDMA transport
4. Restart Glusterd service

Actual results:
Service fails with segmentation fault and core dumps

Expected results:
Service starts successfully

Additional info:
Debug log attached
Core dump on this link: https://drive.google.com/file/d/10TNUtnTjpXGe1AaJzW4CAg9dTAe6hX_U/view?usp=sharing

Comment 1 ryan 2019-06-24 11:53:38 UTC
After some more testing, I've found that:
- Issue goes away if the MLNX_OFED package is uninstalled
- Issue exists even if all Gluster volumes are set to TCP transport

Comment 2 Amar Tumballi 2019-07-15 07:24:19 UTC
Hi Ryan,

RDMA will not be supported from glusterfs-8.0, and hence marking this bug as
WONTFIX/EOL.

(ref: https://review.gluster.org/23033)