CONTAP-448985: ONTAP cluster node reboots unexpectedly
Issue
ONTAP cluster node reboots unexpectedly:PANIC: prod/common/wafl/free_cache.c:371: Assertion failure. in SK process sd_reformat_main on release 9.16.1P1 (C)
- When SE (storage efficient) data is replicated from one platform to another, a conversion from the source compression algorithm to the algorithm supported on the destination is sometimes necessary. This conversion allows the destination to maintain storage efficiency of the replicated data, and is the reformat referred to in the Panic string.
- As part of this reformatting, memory is pre-allocated for this operation.
- During this pre-allocation we should not context switch.
- In this scenario, we inadvertently had a context switch while the preallocation was in progress.
- The Assertion and associated panic happened when the pre-allocation slept on the initial CPU, and then woke on a different CPU.
- This problem is being addressed in future versions of ONTAP.