Reset Search
 

 

Article

FN-2019-439, SFP breakout subinterfaces may be incorrectly set to administratively down in NOS 7.4.0

« Go Back

Information

 
Notice Summary
A VDX6740, VDX6740T, or VDX6740T-1G with SFP breakout settings and a 10G SFP in a Mellanox QSFP slot adapter inserted into any of slots 49 to 52 may set the TenGigabitEthernet subinterface to administratively down after upgrade to NOS 7.4.0.
Background
NOS 7.4.0 includes software changes affecting handling of hardware interrupt request signals. These changes have led to increased sensitivity to interrupts from 40G QSFP slots holding 10G SFP optics in Mellanox QSFP slot adapters. A side effect of this increased sensitivity is that TenGigabitEthernet subinterfaces created with “sfp breakout” that depend on 10G SFP optics in QSFP slot adapters can be set to administratively down.
Impact
Users of 10G SFP optics in Mellanox QSFP slot adapters in slots 49 to 52 of VDX6740, VDX6740T, and VDX6740T-1G switches should not upgrade to NOS 7.4.0 until 7.4.0a patch and later are available. Users of 10G SFP optics in Mellanox QSFP slot adapters who have already upgraded to NOS 7.4.0 should plan a maintenance window where they attempt “shutdown” and then “no shutdown” or the equivalent on devices connected to their TenGigabitEthernet subinterfaces. If their VDX subinterfaces go administratively down, they should try bringing the interfaces back up with “no shutdown”. If they cannot bring the subinterfaces interfaces back up with “no shutdown” alone, they should downgrade to NOS 7.3.0a and then execute “no shutdown” to bring the interfaces back up.
Products Affected
VDX6740, VDX6740T, VDX6740T-1G When Using Mellanox QSFP Adapters with 10G SFP
Software Affected
NOS 7.4.0
Symptoms
A TenGigabitEthernet subinterface that has experienced this issue will display the following messages when you try to bring it up before reloading or after “ha failover”:
VDX1(conf-if-te-1/0/51:1)# no shutdown
2019/07/11-15:30:51, [NSM-1019], 719, SW/0 | Active | DCE, INFO, VDX1,  Interface TenGigabitEthernet 1/0/51:1 is administratively up.
2019/07/11-15:30:51, [CBR-1042], 720, SW/0 | Active, ERROR, VDX1, UPSM [OID 0x43014030] (114) (UP04): Port hard fault, reason code = 15. Port is now offline. No shut to bring back port online
2019/07/11-15:30:51, [NSM-1020], 721, SW/0 | Active | DCE, INFO, VDX1,  Interface TenGigabitEthernet 1/0/51:1 is administratively down.

Immediately after a reload, an affected subinterface will link up one time:
 
VDX1(conf-if-te-1/0/51:1)# no shutdown
2019/07/11-15:46:48, [LOG-1000], 849, SW/1 | Standby, INFO, VDX6740T, Previous message has repeated 1 times.
2019/07/11-15:46:48, [NSM-1019], 850, SW/0 | Active | DCE, INFO, VDX1,  Interface TenGigabitEthernet 1/0/51:1 is administratively up.
2019/07/11-15:46:49, [NSM-1001], 851, SW/0 | Active | DCE, INFO, VDX1,  Interface TenGigabitEthernet 1/0/51:1 is online.


If the link goes down due to an event such as disabling and re-enabling from the other end, the condition will reoccur.
VDX1#
2019/07/11-15:51:11, [ONMD-1008], 861, SW/0 | Active | DCE, INFO, VDX1, Received shutdown LLDP PDUs with TTL=0 on TenGigabitEthernet 1/0/51:1 from MAC address mac address cc4e.2493.a100.
2019/07/11-15:51:11, [NSM-1002], 862, SW/0 | Active | DCE, INFO, VDX1,  Interface TenGigabitEthernet 1/0/51:1 is protocol down.
2019/07/11-15:51:11, [NSM-1003], 863, SW/0 | Active | DCE, INFO, VDX1,  Interface TenGigabitEthernet 1/0/51:1 is link down.
2019/07/11-15:51:11, [CBR-1042], 864, SW/0 | Active, ERROR, VDX1, UPSM [OID 0x43014030] (114) (UP04): Port hard fault, reason code = 15. Port is now offline. No shut to bring back port online
2019/07/11-15:51:12, [NSM-1020], 865, SW/0 | Active | DCE, INFO, VDX1,  Interface TenGigabitEthernet 1/0/51:1 is administratively down.

A reload or “ha failover” will be necessary to bring the link back up again.
Workaround
As a temporary workaround, reloading or executing “ha failover” will allow “no shutdown” to be executed against the TenGigabitEthernet subinterface and bring it up exactly once. If the subinterface goes down again due to an event on the link or on the other end of the link, the subinterface will be set to administratively down again. Another reload or “ha failover” will be necessary to bring the subinterface back up.
Solution
Upgrade to software patch NOS 7.4.0a or later (Defect NOS-67192), or downgrade to a NOS version prior to NOS 7.4.0. If downgrading to NOS 7.3.0a, you can use the coldboot option to preserve the running config and prevent this issue from happening.

Feedback

 

Was this article helpful?


   

Feedback

Please tell us how we can make this article more useful.

Characters Remaining: 255