01-19-2020 11:31 PM
I am looking for a document explaining of different stages during booting of RSP A9K-RSP-880?
01-20-2020 04:07 PM
These are the typical states you will see, additional states like XR PREP, BRINGDOWN, etc are special states.
XR PREP specifically being a state in which the OS is synced from the active card in the chassis to the one booting, that card will then reload to use the right image and flow through the regular states.
0. NOT-PRESENT
This is technically the initial state. The node is not present in the system, waiting for OIRd. A problem here indicates that the card is not properly seated, not getting powered, or another low-level issue.
1. PRESENT
Upon a node being powered-on and Shelf Manager detecting the presence of a card using the OIRd register we changes states from: NOT_PRESENT --> PRESENT
Shelf Manager is pending the receipt of a boot request.
This state has a 300s timer (TimeoutA) to send a boot request before rebooting the node.
2. ROMMON
Once Shelf Manager receives a boot request and the card type is known, we again change states: (PRESENT --> ROMMON)
3. MBI-BOOT
Shelf Manager validates the boot request and sends a boot reply. In order to validate the boot image Shelf Manager will call the mbimgr API.
If the node's bootflash does not have the boot image the image will be downloaded via the Control Ethernet TFTP server. The Node's state is again changed: (ROMMON --> MBI-BOOTING)
This state has a 500s timer (TimeoutB) to send MBI-HELLO HBs before rebooting the node.
4. MBI-RUNNING
The mbi-hello process is initialized during MBI Band by the init process and starts sending MBI-HELLO messages to shelf manager. Upon reception of the first MBI-HELLO message (MBI-HELLO HB) shelf manager will transition the node state from MBI-BOOT to MBI-RUN. Shelf manager stops the MBI-HELLO timer and starts the XR-HELLO timer (15 minute timeout value).
When Shelf manager receives the first MBI-HELLO, the node state changes to: RUNNING_MBI.
This state has a 900s timer (TimeoutD) to send hbagent HBs (XR-HELLO) before going to XR-FAIL.
5. XR-RUN
The hbagent process is initalized by sysmgr process and starts the hand-off operation with mbi-hello process. After the hand-off is completed the mbi-hello process will exit and hbagent process will start sending XR-HELLO messages to shelf manager. Upon reception of the first XR-HELLO message shelfmgr will transition the node state from MBI-RUNNING to XR-RUN. Shelf manager stops the XR-HELLO timer. Now, the node begins sending heart beat messages every second to keep it in RUNNING_ENA (IOS-XR RUN)
Note there is a lot more that happens, but that is the basics.
Sam
01-20-2020 10:05 PM
Thanks for the prompt response..
also can we sync the software from rsp440 to rsp880?
01-21-2020 02:28 PM
Here's a Cisco post regarding RSP-440 to RSP-880 migration:
01-21-2020 10:08 PM
what stage is it?
Discover and save your favorite ideas. Come back to expert answers, step-by-step guides, recent topics, and more.
New here? Get started with these tips. How to use Community New member guide