Connectivity Dropping on 3750X 4 Times Each Day Around Same Times. Need help.

[martinezaw]

I have a 7 switch stack...

I keep getting gaps like the one below in my monitoring system 4 times a day. It happens around the same time a day. I checked the I/O at the time the alert comes (I am still able to log in funny enough) and I don't see anything out of the ordinary. Could this be a bug of some sort? Anyone experience something like this in a stack before? I have been on the switch when the alert comes and it does seem to lag a bit but I don't see anything looking like it is being overtaxed. Nothing shows up in logging either of significance.

Switch Ports Model              SW Version            SW Image                 

------ ----- -----              ----------            ----------               

     1 54    WS-C3750X-48       15.0(1)SE3            C3750E-UNIVERSALK9-M     

     2 54    WS-C3750X-48       15.0(1)SE3            C3750E-UNIVERSALK9-M     

     3 54    WS-C3750X-48       15.0(1)SE3            C3750E-UNIVERSALK9-M     

*    4 54    WS-C3750X-48P      15.0(1)SE3            C3750E-UNIVERSALK9-M     

     5 54    WS-C3750X-48       15.0(1)SE3            C3750E-UNIVERSALK9-M     

     6 54    WS-C3750X-48P      15.0(1)SE3            C3750E-UNIVERSALK9-M     

     7 54    WS-C3750X-48P      15.0(1)SE3            C3750E-UNIVERSALK9-M     

[Reza Sharifi]

Do the users or hosts actually lose connectivity when this happens or you just get an alert and respond to it.

If the users don't lose connectivity and there is no complain than this may be a bug in the management software you are running.

Also, check for any STP issues.

HTH

[martinezaw]

Haven't heard of any complaints from others. I haven't being doing any connectivity tests myself though.

Several IDF switches uplink to this switch and those switches don't have any monitoring gaps... Yet the on-site IDF switches have to go through the problem switch to get to the monitoring server. 

[Leo Laohoo]

1. 15.0(1)SE3 ... that is one royally f*cked IOS version right there.
2. A stack of 7 switches ... And what is the uptime?

[martinezaw]

What is wrong with that version? Is there some known issue? I know it is old but anything in particular?

Uptime is over 6 weeks.

It's been doing this for months and months.

[Leo Laohoo]

---

@martinezaw wrote:

What is wrong with that version? Is there some known issue? I know it is old but anything in particular?

---

It is not just old but Cisco has deemed this train as "deferred".  

Another thing, I tested this version and I was not really impressed by it.  Switches running this IOS crashed very regularly.  So I rolled back to 12.2(55)SE train.

[martinezaw]

Ok, when I get a chance I will probably try to see if changing the IOS version helps.

[martinezaw]

Also any issues or caveats downgrading from 15.0 to 12.2?

[Leo Laohoo]

Other than it will take about 45 minutes because of the microcode upgrade? I don't see any.
You could also try 15.0(2)SE train. That's also stable.

[martinezaw]

Ok, I upgraded the code on the switch stack to 15.0.2 SE9 and still the issue is occurring with the dropped monitoring data.

Could it be the size of such a large stack? It is our largest stack in the district with 7 switches.

[Leo Laohoo]

So when it "drops", do you see any "log" entries?
Did the stack reboot?
Did the up link drop?