Bug ID 751448: TMM, ZebOS, and Linux routing table may lose dynamic routes on a tmm restart

Last Modified: May 29, 2024

Affected Product(s):
BIG-IP TMOS(all modules)

Known Affected Versions:
14.1.0, 14.1.0.1, 14.1.0.2, 14.1.0.3, 14.1.0.5, 14.1.0.6, 14.1.2, 14.1.2.1, 14.1.2.2, 14.1.2.3, 14.1.2.4, 14.1.2.5, 14.1.2.6, 14.1.2.7, 14.1.2.8, 14.1.3, 14.1.3.1

Fixed In:
15.0.0, 14.1.4

Opened: Nov 29, 2018

Severity: 3-Major

Symptoms

There are three major routing participants on a BIG-IP system: TMM, ZebOS, and Linux routing tables. Each of them replicates routes between the other. The 'bigstart restart tmm' command restarts tmm, and a part of the restart process is to mark VLAN interfaces DOWN and later UP. Another part the same process is to restart the ZebOS daemons. There is a race condition between these two events, so the following might happen: 1) tmm marks interface named vlan1 as DOWN, and a bit later marks as UP, but not UP and RUNNING. 2) The ZebOS daemons are restarted and ready to update interface status. They request a current status and mark interface UP, not UP and RUNNING. 3) tmm is fully restarted and marks vlan1 UP and RUNNING. 4) The ZebOS daemons reject dynamic routes because interface vlan1 is UP, but not RUNNING.

Impact

Traffic which relays on dynamic routes is interrupted. Because this is a race condition, it depends on configuration and timing.

Conditions

- BIG-IP Virtual Edition (VE). - Dynamic routing is configured and there is a decide with some dynamic routes. - You run the 'bigstart restart tmm' command.

Workaround

Restart tmrouted daemon using the following command: bigstart restart tmrouted

Fix Information

Dynamic routes are not rejected and are successfully inserted into routing tables.

Behavior Change

Guides & references

K10134038: F5 Bug Tracker Filter Names and Tips