To replace a faulty node canister with a new one received
from CRU or FRU stock, use this procedure. When replacing a node canister,
aim to maximize drive and system availability by maintaining one online
node in the control enclosure containing the faulty node canister.
If you cannot maintain at least one node canister online in the system,
then you might need to follow the system recovery procedure after
replacing the faulty node canister.
- Follow Procedure: Removing a node canister to remove the
faulty node canister.
- Remove the lid of the faulty canister, as described in Procedure: Removing and replacing lid of a node canister. Do the same to the replacement canister.
- Some components inside the faulty node canister must be
transferred to the replacement canister. Transfer each of the following:
- Replace the lid of the faulty canister and the lid of the
replacement canister. Refer to Procedure: Removing and replacing lid of a node canister.
- Open the release levers of the replacement canister.
- Push the replacement canister into the slot of the enclosure
that the faulty canister was removed from, until it stops.
- Finish inserting the replacement canister by closing both
release levers and pushing both orange locks into place.
- The canister is correctly installed when the rear face
of the canister is flush with the rear edge of the enclosure. If the
enclosure is powered and the canister is correctly installed, the
canister starts automatically. Repeat step 5,
if the canister is not correctly installed.
- Reinsert the data cables into the ports that they were
originally connected.
- If no node canisters are online, the system is not online.
To recover the system in the case when no node canisters are online,
see Recover system procedure.
- If only the replacement node is in service state with node
error 503, apply Procedure: Rescuing node canister software from another node (node rescue) to rescue the replacement node canister.
- When the node canister is powered up, it is automatically
added to the system and the system automatically ensures that the
machine code version on the new canister matches that of the other
node canister in the control enclosure. This is reflected in the system
event log.
- When the canister is back online, check the event log for
any new events that might indicate a problem with the reassembly.