cancel
Showing results for 
Search instead for 
Did you mean: 
Create Post
Level 9

NPM 12 Stack Data Ring error

Jump to solution

Anyone else experiencing this error with NPM 12? The power stack data displays fine, I just get this nasty error when I add the "Stack Data Ring":

Unable to construct the data ring diagram because the information required to show the diagram is missing or is not complete. Make sure the device is configured properly.

This is the case for every single stack we have.

0 Kudos
1 Solution

Dan,

Just for giggles and to compare with what I've seen with "ghost" switches. Check how many switches are provisioned since I've seen that if there is a member provisioned but not in a "ready" state that it throws Solarwinds a curveball in data ring reporting. I'll post an update once I remove the "ghost" provisioned switches.

View solution in original post

38 Replies
Level 13

Is there a way to not poll Stack-power but keep the data ring?  We have multiple dual-stacked 3650's. Those switches do not have the powerstack capability

I have multiple log entries where we see that SolarWinds tries to gather the information:

     2016-10-31 17:23:46,099 [STP SmartThreadPool Thread #199568] WARN  SolarWinds.NPM.Pollers.Technology.SwitchStack.CiscoSwitchStackPollerHelper - 10.xx.xx.xx - StackPower data was not discovered

How could we clean this up?

0 Kudos
Level 8

     Actually, I'm seeing this problem on a 3850 running 03.07.04E (Paired nodes, no Stack data) and I'm missing one out of a four-node setup of 3850's running the same version.  Has anyone seen similar of late?  I've ensured that the stack poller is both compatible and applied, and that all stack related elements are being polled, but no joy.

tr,

Are the switches/stack ports reporting all up on the switch side? Check provisioning as well.

"sh switch" and "sh switch stack-ports" should let you know that they are all connected and provisioned properly.

(results: 3850 pair w/ 3.0.7.04E)

The Switch Stack Members pane provides data, which matches the sh switch results 1 for 1.  (hw version v02)

Sh switch stack-ports show both ports for both switches as OK.

Resources view in SW shows that both stackports are listed and polled, as well as the related StackSub resources.

(results: 3850 quad stack)

sh switch has leaves 1-3 all good to go and matching each other (hw v06), but the 4th switch is version 0.  Possible problem there, even though all are listed as ready?

Sh switch stack-ports show both ports for all switches as OK.

Resources view in SW shows all four sets of stackports and StackSub resources as present and polling.

The 4th switch being version 0 may just be a reboot to fix as long as it seems to be working normally. It could however impact your data stack reporting. Can you post the output of your "sh switch" cmd on the quad stack just so I can lay eyes on it ?

0 Kudos

Dan,

I finally got the error message you are receiving but mine is on a stack of 9 3850s running 03.06.03E. I've checked the poller, but depending on your settings if the poller isn't assigned to that device (atleast in my setup) then the stack tab doesn't even show up. One thing I did find strange as I was investigating this was one of my stack rings is apparently not reporting/using power...more issues yay See below

#sh switch detail

Switch/Stack Mac Address : dceb.9440.4600 - Local Mac Address

Mac persistency wait time: Indefinite

                                             H/W   Current

Switch#   Role    Mac Address     Priority Version  State

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

*1       Active   dceb.9440.4600     15     V04     Ready

2       Member   dceb.9440.4680     1      V04     Ready

3       Member   b07d.47b7.2380     1      V04     Ready

4       Member   bc16.f503.f800     1      V04     Ready

5       Member   dceb.9420.fc80     1      V04     Ready

6       Member   2834.a2fc.8680     1      V04     Ready

7       Member   b07d.47e0.9d80     1      V04     Ready

8       Member   b07d.47a7.cc80     1      V04     Ready

9       Standby  dceb.9440.5080     14     V04     Ready

         Stack Port Status             Neighbors

Switch#  Port 1     Port 2           Port 1   Port 2

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

  1         OK         OK               2        6

  2         OK         OK               3        1

  3         OK         OK               4        2

  4         OK         OK               5        3

  5         OK         OK               9        4

  6         OK         OK               1        7

  7         OK         OK               6        8

  8         OK         OK               7        9

  9         OK         OK               8        5

stackdata.PNGstackpower.PNG

sh stack-power

Power Stack           Stack   Stack    Total   Rsvd    Alloc   Unused  Num  Num

Name                  Mode    Topolgy  Pwr(W)  Pwr(W)  Pwr(W)  Pwr(W)  SW   PS

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

Powerstack-1          SP-PS   Ring     4290    42      4248    0       3    6

Powerstack-11         SP-PS   Ring     4290    34      4256    0       3    6

Powerstack-12         SP-PS   Ring     0       0       0       0       0    0

beardedgeeknc, maybe problem is your stack cabling/configuration ?...

yours...

stack_1.jpg

Mine...

stack_2.jpg

dstj,

Thanks for pointing that out, but it's how I set it up since the switches aren't located in the same rack but next to each other. 5 in 1 rack/ 4 in the other.  I don't think that would cause the ring to not report correctly to solarwinds though. You should be able to cable them up in any way and then get a very confusing stack ring

0 Kudos

generally... check out Erdos - Online Graphviz Viewer

you can convert the output of the stack detail into a dot-language specification like this:

digraph stack {

n1->n2;n1->n6;

n2->n3;n2->n1;

n3->n4;n3->n2;

n4->n5;n4->n3;

n5->n9;n5->n4;

n6->n1;n6->n7;

n7->n6;n7->n8;

n8->n7;n8->n9;

n9->n8;n9->n5;

}

if you pick a circular layout (circo) then you end up with a picture like this (which looks good to me)

pastedImage_1.png

you're probably right.  Any chance of going to version 3.6.4.E ?

A few of us on this thread are running it without issue.

0 Kudos

Sadly no. I must wait on my Corporate overseers to finish bug scrubbing 3.6.4E and with some of the downsizing up there recently...no clue when that will or ever be done.

0 Kudos
Level 8

I have dozens of stacks, from 2 to 8 switches.  They all report the data and power stack fine, and draw the power stack, but none of them draw the data stack ring.  Same message as above  Running a 3.6 ver

0 Kudos

I figured this one out.. Apparently you have 2 or more switches in the data stack, it can't draw out the diagram..... I hope I'm wrong!

Try this as a test: Login to solar winds and check one of your 2 switch stacks. The data diagram shows up fine with only 2 switches, but when you have more than 2 it does not seam to work

I will post a pic of it working and not working.

3850's running 3.6.4.E are good:

pastedImage_0.png

2860X stacks running 15.2.4.E1 are cool:

pastedImage_0.png

Dan,

I've been seeing some issues in the reporting of power/data stack rings. I noticed I had some "ghost" provisions were I stacked switches for an IOS upgrade, but later divided them up to deploy in different locations. I've attached the data ring for a stack of 7 3850s I have. I'm getting the same issue with certain stacks. Will update once I have some more time to troubleshoot.

Running 03.06.03E

stack1.PNG

This is more of an oversight on my part, but I had stacks that were not reporting data rings on SW, but reported power stack ring. Turns out I had "ghost" switches provisioned when I had upgraded a stack and divided them later. After unprovisioning the switches and rebooting. SW now reports correctly. Just a heads up from an unobservant net admin

There are some guidelines for building Cisco stacks that might improve your experience:

pastedImage_2.png

0 Kudos

I appreciate the references which are helpful, but I've got multiple stacks all configured similarly that consist of a varying number of  switches with some reporting and some not. It seems like certain stacks are reporting accurately and some are not.

I bet if you opened a TAC case, Cisco would tell you to upgrade the IOS.  A pretty standard response . . .

Is that recommendation from Cisco documentation ?

0 Kudos