[LU-16181] validate_lod_and_idx() message is missing information

Related — Thu, 22 Sep 2022 21:26:08 +0000

We've seen errors like "validate_lod_and_idx() fs-MDT0001: bad idx 103 of 32" in cases where there are sparse OST index numbers (e.g. OST index 0-15, 100-115 in use).

Firstly, the message doesn't print anything about the actual file FID that is having this problem, which should be done in lod_initialize_objects() if validate_lod_and_idx() returns an error.

Secondly, the problem itself shouldn't happen. I suspect it relates to the sparse vs. dense OST index being checked. ltd_tgts_size is the size of the ltd_tgt_idx[] array, but not necessarily the maximum index in the array. lod_initialize_objects() is calling this function with l_ost_idx, which appears to be correct ("/* OST index in LOV (lov_tgt_desc->tgts) */" but is clearly not in some cases.

Whamcloud Community JIRA

[LU-16181] validate_lod_and_idx() message is missing information