I pushed a fix (which is not heavily tested) to wip-8937. The problem seem to have been triggered by the fix to #8928. One thing that I noticed was that previously max_chunk_size at that point was actually used as the minimum write size (how appropriate). It's used elsewhere as the max size for data drained off the front end. Basically the fix for #8928 now makes max_chunk_size to limit the rados writes to no more than that. Since up until now we weren't doing it, then I'm not too sure that this change won't have significant performance implications. The options that we have at this point is: - keep it as it is now, max_chunk_size limits rados writes - revert it to the old behavior, but add some code to handle the pool alignment when needed
In any case, I don't think this is ready to go into firefly. Yehuda -- To unsubscribe from this list: send the line "unsubscribe ceph-devel" in the body of a message to [email protected] More majordomo info at http://vger.kernel.org/majordomo-info.html
