Bug ID 1024421: At failover, ePVA flush leads to clock advancing and MPI timeout messages in TMM log

Last Modified: Aug 25, 2022

Bug Tracker

Affected Product:  See more info
BIG-IP All(all modules)

Known Affected Versions:
15.1.0, 15.1.0.1, 15.1.0.2, 15.1.0.3, 15.1.0.4, 15.1.0.5, 15.1.1, 15.1.2, 15.1.2.1, 15.1.3

Fixed In:
15.1.3.1

Opened: Jun 09, 2021
Severity: 3-Major

Symptoms

TMM log shows clock advancing and MPI timeout messages: notice slot1 MPI stream: connection to node aborted for reason: TCP RST from remote system (tcp.c:5201) notice slot1 tmm[42900]: 01010029:5: Clock advanced by 6320 ticks

Impact

Upstream switch could receive flow response from both active and standby units and cause a traffic disturbance.

Conditions

-- pva.standby.flush DB key set to 1 (enabled). The default is 0. -- Processing high traffic volume for some time

Workaround

None

Fix Information

Modified the "Pva.Standby.Flush" DB key to take two new values ("2" and "3"). This DB key defines actions that the system takes when a traffic-group goes standby. The values of this DB key are now: -- 0: do nothing (the default) -- 1: evict all ePVA accelerated flows for all traffic-groups -- 2: inform the ePVA to stop processing traffic destined for the MAC masquerade address for this traffic-group -- 3: perform both of the above actions (evict all ePVA accelerated flows, and inform the ePVA to stop processing traffic for the MAC masquerade address)

Behavior Change

The DB variable Pva.Standby.Flush accepts two new values ("2" and "3"). This DB key defines actions that the system takes when a traffic-group goes standby. The values of this DB key are now: -- 0: do nothing (the default) -- 1: evict all ePVA accelerated flows for all traffic-groups -- 2: inform the ePVA to stop processing traffic destined for the MAC masquerade address for this traffic-group -- 3: perform both of the above actions (evict all ePVA accelerated flows, and inform the ePVA to stop processing traffic for the MAC masquerade address)