Check Point Dynamic Balancing – Historical Interface-to-CPU Affinity

[mcvarona]

Hi CheckMates,

I’d like to ask if there is a command or method to view the historical CPU affinity of a specific interface over time specifically, which CPU was assigned to a particular interface (e.g., `ethX`) before an issue occurred.

[Expert@MyGW:0]# fw ctl affinity -l
eth0: CPU 0
eth1: CPU 0
eth2: CPU 0

Issue observed:
- `bond1.X` – No latency to the next-hop gateway
- `bond2.X` – No latency to the next-hop gateway
- `bond3.X` – High latency to the next-hop gateway (100–500 ms)
- `bond3.Y` – High latency to the next-hop gateway (100–500 ms)

- `bond3.Z` – High latency to the next-hop gateway (100–500 ms)

Member Interfaces are up, no drops.
Speed and Duplex are matching with peer
Just the specific bond facing to switch had an issue.

We identified that Dynamic Balancing was disabled. Since the symptoms matched those described in SK163815, we followed the recommendation to restart Dynamic Balancing, which addressed the issue.

However, we would like to investigate why Dynamic Balancing was disabled in the first place.

According to SK163815, this can occur when configuration changes are made while Dynamic Balancing is active, including:
- Changes to CoreXL Firewall instance affinity, instance state, or the number of instances.
- Changes to Multi-Queue affinity or mode, or to the number of RxTx queue weights.

My questions are:
1. Is there a command or log that can show the historical CPU affinity assigned to a specific interface?
2. Is it possible to identify what configuration change caused Dynamic Balancing to be disabled?
3. Are there specific logs or troubleshooting commands that can help identify the change and its source?

Any guidance or recommendations would be appreciated.

Thank you!

[mcvarona]

[dsd 27426 4125788096]@FW2[12 Oct 20:42:57] ds_dmd_start_wt: Adding WT on DMD 0
[dsd 27426 4125788096]@FW2[12 Oct 20:42:57] ds_dmd_start_wt: Notifying DMD MGR to start WT 4 on cpu 9
[dsd 27426 4125788096]@FW2[12 Oct 20:42:57] ds_post_change_updates_do: called from context: wt
[dsd 27426 4125788096]@FW2[12 Oct 20:42:57] recover_record_runtime_state: Recording runtime state
[dsd 27426 4125788096]@FW2[12 Oct 20:42:59] FWs filtered average: 7.1 | SNDs filtered average: 19.3
[dsd 27426 4125788096]@FW2[12 Oct 20:43:41] ds_trigger_mq_mng: about to run dynamic_split -o verify_state
[dsd 27426 4125788096]@FW2[12 Oct 20:43:46] ds_verify_state_do: verify state
compare_interface_state: mismatching q-weight for irq 3 on cpu 5, inf1 has 1, inf2 has 0

compare_interfaces_state: state mismatch for interface eth1-01
[dsd 27426 4125788096]@FW2[12 Oct 20:43:47] ds_verify_state: state mismatch detected (reason: interface_state)
ds_recover_state: Will not recover state from context 'State recovery'. recover attempt 1 exceeds limit 0
[dsd 27426 4125788096]@FW2[12 Oct 20:43:47] ds_handle_state_verification_failure: Initiating shut-down due to state mismatch detection (reason interface_state)
[dsd 27426 4125788096]@FW2[12 Oct 20:43:49] dsd_sig_stop_handler: Gracefully stopping DS, caught sig 3a
[dsd 27426 4125788096]@FW2[12 Oct 20:44:22] ds_fini_and_exit: called
[dsd 27426 4125788096]@FW2[12 Oct 20:44:22] ds_fini: called

[Ilya_Yusupov]

Hi @mcvarona ,

You can review logs under $FWDIR/log/dynamic_split.elg, for example if the coreXL configuration is none default this is what you see in the log:

[Thu Oct 15 13:14:02 IDT 2026] INITIALIZATION=1 DEFAULT_BOOT_STATUS=1 STD_DIV_MULTIPLAYER=1 OFF_BY_DEFAULT_ON_VSX=0 MACHINE_CONFIG_OK=1 NUMA_SPREADING_SORT=0 MONITOR_MODE=0 USER_TOUCHED_DEF_BOOT_STATUS=0 ALLOW_NON_DEF_INST=0 LOG_LEVEL=0 CLUSTER_MODE=1 AUTOMATION_MODE=0 AFFINITY_PREFER_SHMEM=1 STATE_VRIFICATION_FAILURE_DETECTED=0 RX_NUM_PERF_MODE=0 VERBOSE_DEBUG=0 DMD_AUTOMATION=0 REQUIRE_CP_APPLIANCE=1 EMERGENCY_CPU_HANDLING_THRESHOLD=50 ALPHA=10 SND_TO_DEFAULT_ON_BOOT=0
[Thu Oct 15 13:14:02 IDT 2026] instances num is not the default
[Thu Oct 15 13:14:02 IDT 2026] Called from startup, not applying settings
[Thu Oct 15 13:14:02 IDT 2026] Handling bad machine state
[Thu Oct 15 13:14:02 IDT 2026] Bad machine config detected after being enabled
[Thu Oct 15 13:14:03 IDT 2026] Set mq_mng to auto on vs 0 for Mgmt eth1-01
[Thu Oct 15 13:14:13 IDT 2026] Unable to turn on Dynamic Balancing due to requirement not being met.

as for MQ you can check under $FWDIR/log/mq_mng.elg which SND's were assigned on each interface.

Thanks,

Ilya