Age | Commit message (Collapse) | Author | Files | Lines | |
---|---|---|---|---|---|
2017-02-12 | Early break if chunks_to_remove becomes emptyliw/forget | Lars Wirzenius | 1 | -0/+2 | |
2017-02-12 | Drop unused import | Lars Wirzenius | 1 | -1/+0 | |
2017-02-12 | Drop a bit of debug logging | Lars Wirzenius | 1 | -2/+0 | |
2017-02-12 | Simplify logging call | Lars Wirzenius | 1 | -2/+2 | |
2017-02-12 | Reduce memory dumping a bit | Lars Wirzenius | 1 | -8/+0 | |
2017-02-12 | Log length of remaining chunks to remove | Lars Wirzenius | 1 | -0/+3 | |
2017-02-12 | Use more generators to save memory | Lars Wirzenius | 1 | -11/+9 | |
This might cost in terms of time. | |||||
2017-02-12 | Change how uniquely-in-remove-gen computation is done | Lars Wirzenius | 1 | -17/+13 | |
Rather than computing a large union of all chunk ids in all genearations tha will remain, we iterate over the chunk ids and remove them from the set of chunk ids in the generation that is getting removed. This should cut down on memory use. | |||||
2017-02-12 | Log as we count files | Lars Wirzenius | 1 | -0/+2 | |
2017-02-12 | Count files in generation | Lars Wirzenius | 1 | -0/+3 | |
2017-02-12 | Attempt to comput chunk ids in a geneartion more efficiently | Lars Wirzenius | 1 | -6/+7 | |
2017-02-12 | Update copyright years; reformat to pacify pep8, pylint | Lars Wirzenius | 1 | -8/+17 | |
2017-02-12 | Rewrite get_generation_chunk_ids | Lars Wirzenius | 1 | -13/+14 | |
2017-02-12 | Add more memory profile debugging | Lars Wirzenius | 1 | -1/+12 | |
2017-02-12 | Add more memory profile debugging | Lars Wirzenius | 1 | -1/+2 | |
2017-02-12 | Add more memory profile debugging | Lars Wirzenius | 1 | -0/+7 | |
2016-07-02 | Drop r/o caches while forgetting to reduce memory use | Lars Wirzenius | 1 | -0/+9 | |
2016-06-25 | Add flush_client method and fix RepositoryClientNotLocked instance | Lars Wirzenius | 1 | -0/+3 | |
This may be used to force any in-memory caches for an open client in a repository to be written out, and then dropped, to reduct memory use. Only green-albatross implements this, for now. | |||||
2016-03-17 | Add RepositoryInterface.get_client_checksum_key | Lars Wirzenius | 1 | -0/+5 | |
2016-03-17 | Store/use checksum algorithm in per-client data | Lars Wirzenius | 1 | -1/+10 | |
2016-03-17 | Do union only once | Lars Wirzenius | 1 | -5/+6 | |
2016-03-13 | Use well-known blob, not raw object, in per-client | Lars Wirzenius | 1 | -8/+7 | |
2016-03-11 | Only clone a DIR obj when actually changing it | Lars Wirzenius | 1 | -7/+14 | |
2015-12-21 | Change GA remove_generations to find unused chunks | Lars Wirzenius | 1 | -2/+14 | |
2015-12-21 | Change remove_generation to return list of chunks | Lars Wirzenius | 1 | -0/+2 | |
This will be useful for actually implementing removal of unused chunks in green-albatross. | |||||
2015-12-06 | Add green-albatross cache settings | Lars Wirzenius | 1 | -3/+14 | |
2015-11-01 | Avoid calling _require_file_exists | Lars Wirzenius | 1 | -2/+5 | |
This should avoid a little overlapping work. | |||||
2015-08-29 | Rename unused variables to _ | Lars Wirzenius | 1 | -4/+4 | |
2015-08-28 | Drop unused method | Lars Wirzenius | 1 | -3/+0 | |
2015-08-20 | Avoid _require_file_exists in get_file_children | Lars Wirzenius | 1 | -3/+8 | |
2015-08-20 | Avoid _require_file_exists in clear_file_chunk_ids | Lars Wirzenius | 1 | -3/+8 | |
2015-08-20 | Avoid calling _require_file_exists in get_file_chunk_ids | Lars Wirzenius | 1 | -3/+8 | |
2015-08-20 | Avoid double check due to call to _require_file_exists | Lars Wirzenius | 1 | -3/+8 | |
Instead, raise the error if it happens when doing the actual work. | |||||
2015-08-19 | Add get_metadata_from_file_keys and use it | Lars Wirzenius | 1 | -21/+36 | |
2015-08-19 | Move set_file_keys_from_metadata further down the stack | Lars Wirzenius | 1 | -28/+41 | |
2015-08-19 | Add set_file_keys_from_metadata to interface | Lars Wirzenius | 1 | -0/+34 | |
2015-08-02 | Move method closer to first use, for clarity | Lars Wirzenius | 1 | -6/+6 | |
2015-08-02 | Lookup generation in dict, not linearly | Lars Wirzenius | 1 | -5/+11 | |
2015-08-02 | Move file-exists checking deeper into call stack | Lars Wirzenius | 1 | -6/+6 | |
2015-08-02 | Inline _require_file_exists in case it's faster | Lars Wirzenius | 1 | -1/+7 | |
2015-07-25 | Drop --ignore-missing-chunks option | Lars Wirzenius | 1 | -1/+1 | |
We want robustness by default, and always. | |||||
2015-07-25 | Add --ignore-missing-chunks option to forget | Lukáš Poláček | 1 | -1/+1 | |
2015-07-19 | Only use one blob store per client | Lars Wirzenius | 1 | -8/+11 | |
2015-07-19 | Split chunk and DIR object cache sizes | Lars Wirzenius | 1 | -0/+1 | |
2015-07-17 | Simplify get_file_children in Green Albatross | Lars Wirzenius | 1 | -3/+12 | |
2015-07-17 | Pack more DIR objects in a bag | Lars Wirzenius | 1 | -1/+1 | |
2015-07-11 | Don't return . as a file | Lars Wirzenius | 1 | -1/+3 | |
2015-07-11 | Drop unnecessary file removal | Lars Wirzenius | 1 | -1/+0 | |
Either it's a new (and empty) parent directory, or the parent directory didn't have the file, when the file was originally added. | |||||
2015-07-11 | Bug fixes | Lars Wirzenius | 1 | -1/+2 | |
2015-07-11 | Change GAFileMetadata to use GATree | Lars Wirzenius | 1 | -50/+148 | |