Last Modified: Dec 19, 2025
Affected Product(s):
BIG-IP LTM
Known Affected Versions:
17.5.1.3, 21.0.0
Opened: Oct 27, 2025 Severity: 3-Major
TMM crashes due to lack of memory to configure subchannels needed for queues in DPDK which ultimately results in a NULL pointer exception. The lack of memory occurs when the product of (number of TMMS)*(number of NICs) becomes very large due to memory footprint each TMM needs to operate so many NICs. In /var/log/tmm: notice hn_nvs_alloc_subchans(): nvs subch alloc failed: 0x2 notice hn_dev_configure(): subchannel configuration failed notice Port5 dev_configure = -5
TMM goes into restart loop and never becomes Active, disrupting traffic.
1) xnet-netvsc driver (HyperV or Azure) 2) (number of TMMs)*(number of NICs) is big; confirmed with 8 TMMs and 4 NICs on Azure F8s v2 instance.
A) Reduce the number of NICs in the environment B) Reduce the number of TMMs by running the following and then restarting with 'bigstart restart tmm' tmsh modify sys db provision.tmmcount value <tmm_count>
None