Support Support Downloads Knowledge Base Case Manager My Juniper Community

Knowledge Base

Search our Knowledge Base sites to find answers to your questions.

Ask All Knowledge Base Sites All Knowledge Base Sites JunosE Defect (KA)Knowledge BaseSecurity AdvisoriesTechnical BulletinsTechnotes Sign in to display secure content and recently viewed articles

Syslog message: %PFE-5: fpc19 PFE 3 Hotbank alloc limit reached

0

0

Article ID: KB35177 KB Last Updated: 23 Nov 2019Version: 2.0
Summary:

The syslog message, '%PFE-5: fpc19 PFE 3 Hotbank alloc limit reached' indicates an event has occurred and does not impact service. It can be safely ignored.

Symptoms:

The following log messages are seen on MX2020 running 17.4R2.

Sep 10 09:45:01 root : %PFE-5: fpc19 PFE 3 Hotbank alloc limit reached.
Sep 11 18:14:32  root : %PFE-5: fpc8 PFE 2 Hotbank alloc limit reached.
Sep 14 07:18:54 root : %PFE-5: fpc9 PFE 3 Hotbank alloc limit reached.
Cause:

The forwarding state resides in the Lookup ASIC memory.  The Lookup ASIC memory in turn has a set of caches that front the physical memory. Although the forwarding states are equally spread across the ASIC memory for the static load (same number of locations per cache), the dynamic load (the portion of  the forwarding state that is actually in use) may not load the caches equally. The ASIC driver constantly monitors the load on each of the caches.  When it detects that one of the caches has a load >= 150% of the average load, that cache is marked as ‘heavy’. If it detects that a given cache is nearing the physical limitation of the ASIC, the cache will be marked as ‘hot bank’.

The driver notifies the application layer (JNH) the mask of the ‘heavy’ and 'hotbank’ caches for a while.  JNH in turn will try to avoid allocating additional memory that is associated with the ‘heavy’ and ‘hotbank’ caches with the goal to achieve more fairly balance the load on the caches. JNH maintains a table (a list) of the memory locations that are ‘disabled’ due to the cache load.  It periodically updates the list based on the information  received from the ASIC driver.

In a case of EA based FPCs like the MPC9, the 'distributed caches' referred to are the CAE:

  • Each memory type (XR/IDM/HMC), is fronted by a small cache (256 DWORDS) called a ‘Cache with Atomic Engine’ or CAE.
  • Its purpose is to minimize main memory accesses, and facilitate atomic operations
  • It helps to improve performance when a ‘fat’ flow accesses a small number of entries at rates faster than the memory can support
  • If a CAE is > 94% this is termed ‘HOT-BANKING’
  •  A miss is recorded whenever the address could not be retrieved from CAE’s internal data buffer and is forwarded to the memory block

 For XM based FPC like the MPC6E, the same applies, but the memory types are XR/IDM/DDR each of which are still fronted by CAE.

Solution:

This message has NOTICE level status and indicates normal but significant condition. There is no impact related to this message and it can be safely ignored.

Please note that there is no one command that gives this calculation, it is deduced from a number of outputs and calculations.

Modification History:
2019-11-23: Updated Cause and Solution sections.
Comment on this article > Affected Products Browse the Knowledge Base for more articles related to these product categories. Select a category to begin.

Getting Up and Running with Junos

Getting Up and Running with Junos Security Alerts and Vulnerabilities Product Alerts and Software Release Notices Problem Report (PR) Search Tool EOL Notices and Bulletins JTAC User Guide Customer Care User Guide Pathfinder SRX High Availability Configurator SRX VPN Configurator Training Courses and Videos End User Licence Agreement Global Search