[ixpmanager] Broken aggregated graphs when removing old Switches

[Mauricio Oviedo]

Hi folks,

On Friday I added a new Switch to replace 2 old Switches. I migrated the active ports by editing them and ponting to the new interfaces. All went well.

However, I didn’t inmmediatly marked the old switches as inactive, and some member ports that were no longer connected were still associated to the old Switches. As a result the member ports and switch MRTG graphs were OK, but the aggregated overall graph did not graph anymore.

After a long weekend I found these 2 posts where I got pointers about deactivating the old switches and do some clean up on the unused ports:

https://www.inex.ie/pipermail/ixpmanager/2021-April/003183.html
https://www.inex.ie/pipermail/ixpmanager/2019-July/002297.html

Here a couple of questions:


  1.  Why the aggregated graph stops working if a Switch or port cannot be polled by SNMP?
  2.  Given that the member and switch graphs were OK and the information exists, is it possible to rebuild the aggregated graph to include the missing data? (Around 30hrs)
  3.  What would be the recommended procedure to remove a Switch without loosing historical graphs? (Right now the switches are inactive, but still in IXP Manager)
  4.  What would be the recommended procedure to upgrade a member with 3x 10G ports in LAG to a 100G port, without loosing historical graphs?

Thanks,
Mauricio

-------------- next part --------------
An HTML attachment was scrubbed...
URL: <https://www.inex.ie/pipermail/ixpmanager/attachments/20240415/84295d15/attachment.htm>