Dell Unity: FSN performance impact when Ethernet Switch reporting Unity Ethernet MAC address is flapping between ports (Dell Correctable)
Summary: The customer is using Unity’s Fail Safe Networking (FSN). Both the Primary Active and Standby interfaces are configured with a two-port Link Aggregation Control Protocol (LACP) bond. The LACP bonded interfaces for the FSN Primary Active are connected to the Primary Active Access switch. The FSN Standby LACP bonded interfaces are connected to the Standby Access switch. However, the customer’s Cisco Access Switches do not support LACP virtual Port Channel (vPC) across switches. ...
Symptoms
Unity Fail Safe Networking (FSN) is experiencing performance degradation whenever the Cisco Access Switches issue these MAC_MOVE alerts.
Sep 2 14:27:52.147 UTC+1: %MAC_MOVE-4-NOTIF: Host xxxx.xxxx.xxa4 in vlan 10 is flapping between port Po1 and port Po2 Sep 2 14:27:52.339 UTC+1: %MAC_MOVE-4-NOTIF: Host xxxx.xxxx.xxa4 in vlan 10 is flapping between port Po2 and port Po1
Cause
OpenSLP is Multicasting updates out all active interfaces, including the Unity standby FSN interface.
This causes the switch's management console to see a duplicate MAC address coming from the standby FSN interface on the standby Access switch which causes the switch to post an alert.
Jul 20 17:32:24.624 UTC+1: %MAC_MOVE-4-NOTIF: Host xxxx.xxxx.xxa4 in vlan 10 is flapping between port Po1 and port Po2
The Ethernet switch must converge the network traffic going to the Unity's Ethernet MAC address, that is moved to the standby Access switch.
There is a delay in the network traffic, while the switches reroute the network traffic to the Standby Access switch port.
However, Unity is still using the FSN Primary Active interface to send out its network traffic.
The Switch sees the Duplicate MAC address on Primary Access Switch and posts another alert.
Jul 20 17:32:45.360 UTC+1: %MAC_MOVE-4-NOTIF: Host xxxx.xxxx.xxa4 in vlan 10 is flapping between port Po2 and port Po1
The Ethernet switch must converge the network traffic going to the Unity's Ethernet MAC address, that is moved to the Primary Access switch.
There is a delay in the network traffic, while the switches reroute the network traffic to the Primary Access switch port.
This pattern continues until OpenSLP is configured to only use the Unity Management network interface.
Resolution
-
Network Update using Management Interface: Once these steps are completed, Unity’s OpenSLP uses its Management Interface to update the network that the SMI-S Provider Service is active.
-
Reapplication Required: If any of Unity’s Storage Processors (SPs) are replaced, re-imaged, or upgraded, these steps must be re-applied.
-
Configuration Update: Once OpenSLP’s configuration is updated, the OpenSLP daemon on the Primary SP must be stopped and restarted to load the new configuration.
-
Stopping OpenSLP daemon on Peer: The OpenSLP daemon on the Peer must be stopped. The Peer might still have the OpenSLP daemon running with the old configuration if Unity's ECOM services were previously failed over.
Warning: This task must be executed by a Senior or Technical Lead who has received Unity training.
Detail query
lsof -n | grep -i slpd
lsof -n -i :427
echo "";echo slptool findsrvs service:service-agent; slptool findsrvs service:service-agent;echo "";echo slptool findsrvtypes; slptool findsrvtypes;echo "";slptool findsrvtypes | awk '{ print "echo slptool findsrvs " $1" ; slptool findsrvs " $1 ";echo """}' | sh
Quick query
lsof -n -i :427
slptool findsrvs service:service-agent
Find the line that starts with
;net.slp.interfaces = 192.168.0.1
Add a newline with the text
net.slp.interfaces = mgmt_vdev
diff /etc/slp.conf /etc/slp.conf-bak 192d191 < net.slp.interfaces = mgmt_vdev egrep -v "^#|^;" /etc/slp.conf |sed '/^\s*$/d' net.slp.interfaces = mgmt._vdev
scp -p /etc/slp.conf* peer:/etc/.
systemctl status slpd systemctl stop slpd systemctl start slpd systemctl status slpd
10. Connect to the peer SP.
ssh peer
smartctl status slpd
smartctl stop slpd smartctl status slpd
exit
14. On the Primary SP, verify the update is working by querying OpenSLP daemon status.
Detail query
lsof -n | grep -i slp
lsof -n -i :427
echo "";echo slptool findsrvs service:service-agent; slptool findsrvs service:service-agent;echo "";echo slptool findsrvtypes; slptool findsrvtypes;echo "";slptool findsrvtypes | awk '{ print "echo slptool findsrvs " $1" ; slptool findsrvs " $1 ";echo """}' | sh
Quick query
lsof -n -i :427
slptool findsrvs service:service-agent
15. The output of:
slptool findsrvs service:service-agent lsof -n -i :427
Loopback Management interface Multicast IP address in the range of 224.0.1.0-224.0.1.225 Multicast IP address for Service Location Protocol (SLP) 239.255.255.253
Jul 20 17:32:45.360 UTC+1: %MAC_MOVE-4-NOTIF: Host xxxx.xxxx.xxa4 in vlan 10 is flapping between port Po2 and port Po1
exit svc_inject -t -e svc_service_shell
18. Verify the Service Shell has expired.
svc_inject -l -a