Introduction
This document describes about reasons for Attach Rejects due to "smgr resource unavailable" under network failures for MME.
Problem
******** show mme-service statistics debug *******
Attach Reject: 985493
IMSI Unknown in HSS: 0 Illegal UE: 803
Illegal ME: 0 EPS Not Allowed: 0
Emergency-services-disabled: 177316
Network Failure: 206590
Session Recovery: 0 SMgr Resource Unavail: 39379
S11 EGTP Connection: 217 Auth Procedure Fail: 0
HSS Unavailable: 21300 HSS Subscription Fail: 59105
PGW Selection Fail: 1013 SGW Selection Fail: 85576
S1 Connection: 0 Congestion Policy: 0
Gateway Unreachable: 0 New Call Restrict: 0
Solution:
Regarding the SMgr Resource Unavail counter, it will increase if the following conditions are met :
- MME receiving Cancel Location Request from HSS before we send the Attach Accept to UE which is resulting in Attach Reject with cause Network Failure.
- increments when an Attach Reject message is sent due to session recovery with a cause code of "Network Failure". we need to check if any
- Cancel Location Request from HSS received for same mme when Update Location Request is sent from MME. (will result in incrementing Sessmgr resources unavailable stat).
To confirm this we need some s1 and s6a traces.
Cancel request might be coming for a different diamproxy which indicates the HSS had not cleared the state for the given subscriber which was implicitly detached after the node reload.
From the specs:
If the Update Location Request is received over the S6a interface, the HSS shall send a Cancel Location Request with a Cancellation-Type of MME_UPDATE_PROCEDURE (CLR; see chapter 7.2.7) to the previous MME (if any) and replace the stored MME-Identity with the received value (the MME-Identity is received within the Origin-Host AVP).
It is an expected behavior to have those peaks of mme-hss-detach(371) after a reload of a node which has not been offloaded.