This document describes how to troubleshoot some of the common issues with UCSM registering with UCS Central
Cisco recommends that you have knowledge of these topics:
The information in this document is based on these software and hardware versions:
The information in this document was created from the devices in a specific lab environment. All of the devices used in this document started with a cleared (default) configuration. If your network is live, ensure that you understand the potential impact of any command.
The troubleshooting is focused on self-signed certificate on UCSM and central and not 3rd party certificates
Please ensure these basic checks are completed:
Central# connect local-mgmt
Central(local-mgmt)# test ucsm-connectivity <ucsm_ip>
Packet capture from UCSM registering successfully with Central Provider


DO NOT unregister the central from UCSM. When you unregister all global service-profiles become local to the UCS domain. It is possible to make a local service-profile global again. However, it is a very complex process and has an impact on the service.
If UCS Manager is registered to a UCS Central and that UCS Manager is being upgraded to 3.1.1, then the UCS Manager goes to registering state and is stuck there.
Too many curl errors observed in the Central DME logs
9603: [WARN][0x27699940][Apr 5 18:00:54.714][write:net] write of 3752 bytes using curl failed, code=7, error: 'Couldn't connect to server', ep: https://10.106.74.195:443/xmlInternal/managed-endpoint
9604: [WARN][0x27699940][Apr 5 18:00:54.714][write:net] non-critical curl write error.
From UCSM DME
[INFO][0x682ffb90][Nov 1 16:05:24.886][sam_sec:check_cert_val] X509_verify_cert_error_string - ok
[INFO][0x682ffb90][Nov 1 16:05:24.886][sam_sec:X509VerifyCert] ErrorMsg:ok ErrorNo:0
[INFO][0x682ffb90][Nov 1 16:05:24.886][app_sam_dme:processKey] something wrong with KR-default certificate, status - 18
The problem could be due to the UCSM using old MDS hash instead of SHA1 for the certificates
[WARN][0x674ffb90][Nov 22 19:11:49.227][net:write] write of 546 bytes using curl failed, code=60, error: 'Peer certificate cannot be authenticated with given CA certificates(SSL certificate problem: self signed certificate)', ep: https://10.106.74.234:443/xmlInternal/service-reg
[INFO][0x674ffb90][Nov 22 19:11:49.227][net:certFailure] certificate is bad for connection to ' https://10.136.58.4:443/xmlInternal/service-reg';
Perform these workaround as it causes the UCS Manager to register successfully to UCS Central and fix the certificate error
The default keyring can be regenerated from the UCS Central CLI under the device profile section.
connect policy-mgr
scope org
scope device-profile
scope security
scope keyring default
set regenerate yes
commit-buffer
If the workaround does not resolve please raise a case with Cisco TAC to validate further

If at any time the UCS Manager have been registered to UCS Central initially at a version of 2.1.3 or below. Then during the upgrade to 3.1.1 the registration problem mentioned is still seen.
For this TAC involvement is needed as UCS 2.1.3 and earlier releases, UCSM does not split certificate. TAC need to rehash the certificate so that creates the right softlinks to the certificate.
The issue is due to database goes out of sync between central and UCS
These errors observed in the resource-manager logs
[WARN][0xbbce9940][Aug 11 10:23:18.194][storeMo:mit_init] SQL error [SQLParamData failure: Error while executing the query (non-fatal);
ERROR: duplicate key value violates unique constraint "InstanceId2DN_dn_key"] stmt [INSERT INTO "InstanceId2DN"("instanceId","dn","className","parent") VALUES (?,?,?,?)]
[INFO][0xbbce9940][Aug 11 10:23:18.194][report:exception_handl] FATAL[3|150] /ramfs/buildsa/150407-104741-rev219791-FCSa/resMgr/sam/src/lib/framework/core/sql/MitDbImpl.cc(1167):storeMo: Failed to connect to database. Transaction aborted.
[INFO][0xbbce9940][Aug 11 10:23:18.201][report:exception_handl] ERROR[3|150] /ramfs/buildsa/150407-104741-rev219791-FCSa/resMgr/sam/src/lib/framework/core/proc/Doer.cc(795):exceptionCB: exception encountered during processing: "Failed to connect to database. Transaction aborted." [150] Failed to connect to database. Transaction aborted.
[INFO][0xbbce9940][Aug 11 10:23:18.201][failedCb:tx] TX FAILED
This is a database sync issue please raise a case with Cisco TAC to validate further


Check the Registration Status
If it shows "lost-visibility" UCS Central cannot be reached on one or more required ports. If UCS Central is using the flash GUI (Flex) these ports need to be open to Central: 443, 80, 843. HTML GUI only requires port 443.
UCSM
/var/sysmgr/sam_logs/pa_setup.log
svc_sam_dme.log files on FI
Central
Svc_dme_reg.log
Troubleshooting commands
Central# connect policy-mgr
Central# scope org
Central# scope device-profile
Central# scope security
Central# Show keyring detail
UCSM# scope system
UCSM# scope security
UCSM# show keyring detail
connect local-mgmt
telnet <Central IP> <port>
^ (Shift+6) ] with no spaces to exit
FSM status
scope system
scope control-ep policy
show fsm status
Central# connect service-reg
Central(service-reg)# show fault
Central(service-reg)# show clients detail
Registered Clients:
ID: 1008
Registered Client IP: 10.106.74.194
Registered Client IPV6: ::
Registered Client Connection Protocol: Ipv4
Registered Client Name: DCN-INDIA-FI-A
Registered Client GUID: e832cfc2-548b-11e4-b8f2-002a6a6f6dc1
Registered Client Version: 2.2(6g)
Registered Client Type: Managed Endpoint
Registered Client Capability: Policy Client Module
Registered Client Last Poll Timestamp: 2016-12-08T12:33:36.417
Registered Client Operational State: Registered
Registered Client Suspend State: Off
Registered Client License State: License Graceperiod
Registered Client grace period used: 33
Registered Client Network Connection State: Connected
Registering Cisco UCS Domains with Cisco UCS Central
| Revision | Publish Date | Comments |
|---|---|---|
1.0 |
21-Dec-2016
|
Initial Release |