| 13 | 13 | * 14:30 CZW: I can come up with no more tests to run to find any new ways the cull can go wrong. I have restarted the cull phase of the shuffle under the ~ipptest/replication pantasks. Initial spot check tests on cXXXXgYYYYo exposures do not show any issues, with the cull process correctly removing the to-be-deprecated/very full host instances first, and correctly stopping cull when any file needs repaired. The initial tests yielded execution times of ~30s, with 10% failures (which ran at ~15s). Using the current loading, and running full-time, this translates to about 16 days. |
