[LU-16389] Lustre 2.12.9 ksocklnd crash with 100+GB ethernet

• Type: Bug
• Resolution: Won't Fix
• Priority: Critical
• Affects Version/s: Lustre 2.12.9
• Labels: ORNL
• Environment: RHEL8 running 2.12.9 which is nearly vanilla. This is a 200GiB ethernet setup.
• Severity: 3
• Assignee: Peter Jones
• Reporter: James A Simmons

Description

Using a nearly plain vanilla 2.12.9 Lustre version we see on our production 100GiB system the following crashes from time to time

kernel:LNetError: 6003:0:(socklnd_cb.c:1985:ksocknal_connect()) ASSERTION( (wanted & (1 << 3)) != 0 ) failed:
 kernel:LNetError: 6003:0:(socklnd_cb.c:1985:ksocknal_connect()) LBUG
kernel:Kernel panic - not syncing: LBUG in interrupt.