Closed shaitan closed 10 years ago
Log level changes are wrong - it will produce too many uninformative logs. What I thought of is error path for eblob_fill_write_control_from_ram() and/or eblob_disk_index_lookup() which would print error message with lookup stats
Added new state of base - EBLOB_MERGE_NEEDED. This state means that blob is already defragmented but its size allow him to be merged with another one. Added skipping such blobs if there is no another blob with which it can be merged. Increased verbosity of some logs inside eblob_disk_index_lookup.