Stacking Catalyst 3850 switch

[LemmYY]

Hi experts,

We have a 3 x 3850 stacked switch in the production.

I have found out that it is not in a full ring stack, there is no stack cable connected on switch 1 and switch 3.

Now, I need to connect a cable to it, I wanted to be 100% sure that this will not result any downtime.

I don't need to turn off the switches to connect the stack cable form switch 1 and 3 right?

Thanks!

[Leo Laohoo]

---

@LemmYY wrote:
I wanted to be 100% sure that this will not result any downtime.

---

If you do not know what you are doing, power it all down, and make sure the stacking cables are secure, before powering it all back up.

[LemmYY]

Hi Leo,

I know what I need to do but I am just not 100% sure what would be the outcome if i just connect the stack cable from switch 1 port 2 to switch 4 port 1.

As i said, the 3 stacked switch are running fine, it's just not in full ring stack. Adding that cable from switch 1 to switch 3 will fix it but I am just not 100% sure if it can be done while all switch are powered on.

Everyone wants to avoid to raise a ticket with a downtime so I just wanted to gather some information from the experts if they have done this scenario before. Otherwise yeah last resort would turn off all the switch.

[Leo Laohoo]

First, let me start by saying that since the introduction of IOS-XE, Cisco has been consistently releasing one buggy release of software after another without any fail. 

Next, I have no idea what firmware the stack is currently on nor do I know what is the uptime.  

Finally, I have never "accidentally" crashed a stack when attaching a return stacking cable.  But this is because I have spent several hours understanding the stack before conducting an exercise like this.  

But since the question is for a "100% guarantee nothing is going to happen", I am going to recommend powering down the entire stack.  

[balaji.bandi]

how did you come across the stack is not correct.

can you post - show switch here to look

also follow below guide clear picture how you need to cables to be connected backside.

https://www.cisco.com/c/en/us/td/docs/switches/lan/catalyst3850/hardware/installation/guide/b_c3850_hig/b_c3850_hig_chapter_010.html

[Martin L]

if u have half ring right now; just connect "missing" cable from switch 1 and switch 3 to get full ring stack;  I have done several of those couple years ago; I am 100% sure it will work, no down time, no crashes, and it can be done while all switch are powered on. Just make sure Master switch has priority of 15, (switch 1 priority 15), and 2nd switch has priority 14. or you can show us show switch output.  they all should have same IOS version since they are running as stack right now.

Regards, ML
**Please Rate All Helpful Responses **

[Joseph W. Doherty]

Not 100% positive with the 3850 series, but I believe (?) Cisco recommends one member of the stack members being connected, with a stack cable, should be powered off before making/breaking such a connection.

[Leo Laohoo]

@balaji.bandi, @Martin L, have any of you seen a stacking cable not s_crewed in properly cause the stack master to crash in a 3850?

I have seen this happen to a 3850.  And I have picture of how buggy IOS-XE is!

3850 stack, 16.12.6

 Look at the timeline when the cable became loose.  Look at it.  From 9am the CPU just spiked.  All this because the stacking cable was/became loose.  The cable cause the stack-mgr process to shoot through the roof!

IMPORTANT:  If people are not sure about IOS-XE and stacking cables, POWER THE STACK DOWN!

[Martin L]

Maybe it dependson ios version, Occasionally you can get unexpected reboot but once stack is working on 1 cable, adding 2nd one should not create any problems.  We had added few of 2nd cables Live in production.  PM has screwed up and has  ordered too few stack cables.

Regards, ML
**Please Rate All Helpful Responses **

[Leo Laohoo]

---

@Martin L wrote:
Maybe it dependson ios version

---

No, it is not. 

I can crash any IOS-XE firmware.  I have a few tricks to trigger a legitimate (and not a test crash) crash.  It.  Is.  Laughably.  Easy.

---

@Martin L wrote:
Occasionally you can get unexpected reboot but once stack is working on 1 cable, adding 2nd one should not create any problems.

---

I have no question about adding (or replacing) a 2nd stacking cable because I know how it behaves.  My response is aimed at those who are not familiar with how buggy IOS-XE is and changing or adding a 2nd stacking cable can bring the stack down if not done right.  

Our NMS server, for instance, will flag us as soon as a stacking cable goes down/up.  And if we find out the stack is IOS-XE, we jump on it before something really bad happens.

[balaji.bandi]

@Leo Laohoo - never come across...good to know this scenario...good we started decommissioning cat 3850 and Migrating to Cat9300.

[Leo Laohoo]

---

@balaji.bandi wrote:
good we started decommissioning cat 3850 and Migrating to Cat9300.

---

But you're still on IOS-XE and IOS-XE crashes easily -- Even a child can do it. 

[LemmYY]

Hi Experts,

Above is the output of stack-ports summary. All switches are working fine all ports are up from the 3 switches. The only issue here is that it is not working as a "full ring" as there is no stack cable connected to switch 1 and switch 3.

I just wanted to add as well, all of the IOS version is the same for the 3 switches as all of the 3 switches are in functional working as a stacked switch. Again the only issue is this is not working as a full ring stack because of the missing cable on switch 1 and switch 3.

[Martin L]

Current cables are ok; left port of from sw 1 to right port on sw 2, and sw 2 left port to sw 3 right port. just make sure  sw 1 is master with priority set to 15; not master due to other factors (boot speed, MAC value) ;  I think command should be show switch detail   if not, change switch priority.

this output should be like this

SWT0001#sh sw det
Switch/Stack Mac Address : 7xxxxxxxx80 - Local Mac Address
Mac persistency wait time: Indefinite
H/W Current
Switch# Role Mac Address Priority Version State
------------------------------------------------------------
*1 Active 7018.xxxx.4c80 15 V07 Ready
2 Member 7018.xxxx.1600 1 V07 Ready
3 Member d4ad.xxxx.ef80 1 V07 Ready
4 Standby 7018.xxx.1100 1 V07 Ready
5 Member d4ad.xxxxe000 1 V07 Ready

Stack Port Status Neighbors
Switch# Port 1 Port 2 Port 1 Port 2
--------------------------------------------------------
1 OK OK 2 5
2 OK OK 3 1
3 OK OK 4 2
4 OK OK 5 3
5 OK OK 1 4

U can do this after hours but we had done it during normal hours without any problems;  in worst case, stack will reboot, which may take upto 5 minutes for ports to be active again.

No need for any fancy graphs or MNS software .

Regards, ML
**Please Rate All Helpful Responses **

[paul driver]

Hello
Arrange downtime with change control - Powedown stack add cable (create full stackring), bring stack it back up -  K.I.S.S.