Kenneth Skovhede
58f415d525
Added some additional checks to make sure we do not create volumes that are over the size limit.
2018-01-31 22:03:37 +01:00
Kenneth Skovhede
ef36a9356b
Rewrote the uploader to pass in the index volume blocks, as that is 10x times faster than creating it from the database.
...
This greatly increases performance because the database read is no longer preventing progress.
2018-01-30 08:34:57 +01:00
Kenneth Skovhede
a38d60f348
Fixed output spamming
2018-01-29 06:57:16 +01:00
Kenneth Skovhede
805a411f8f
Fixed a few minor issues with the new approach, and rewrote the zip file packaging to allow packing more data into each volume.
2018-01-28 12:29:02 +01:00
Dan Stahr
31ac40083a
Prevent temp files in VolumeWriterBase from being deleted prematurely.
...
The fix is hacky and the root cause should be investigated properly.
2017-08-13 11:58:39 +01:00
Dan Stahr
a0d63bf003
Reverted unrelated changes
2017-08-12 14:20:35 +01:00
Dan Stahr
4d98960e75
Fixed empty block handling.
...
Previously, empty files were associasted with empty blocks. These blocks were added to volumes which could result in empty, but valid volumes being garbage collected and valid file entries deleted.
This change changes the logic so that empty blocks are not stored at all.
2017-08-12 14:14:23 +01:00
Kenneth Skovhede
a3f2b39d23
Implemented handling of pause/stop/abort in the concurrent code.
...
Implemented the dry-run feature for backups.
2016-02-29 21:52:11 +01:00
Kenneth Skovhede
537fe20838
Added a lot of comments.
...
Fixed a few issues that caused parallel uploads.
Fixed the use of the encryption module so the encryption module is not re-used for multiple jobs.
2016-02-25 19:15:51 +01:00
Kenneth Skovhede
20f7b42945
Fixed channels to use type-safe initializations.
...
Updated logging channel to use a wrapper for later early-filter optimizations.
Various bug fixes and optimizations.
2016-02-24 23:48:42 +01:00
Kenneth Skovhede
3ff2f1eb98
The concurrent processing now seems to work, at least in the base case.
...
Rewrote the handling of index volumes to create them on-demand instead of in-advance.
This makes it much simpler to handle upload failures.
2016-02-22 21:27:12 +01:00
Kenneth Skovhede
8abdf87188
More cleanup and fixes for the concurrent processing
2016-02-20 15:06:29 +01:00
Kenneth Skovhede
ec877696b0
More work on concurrent processing
2016-02-18 16:46:53 +01:00
Kenneth Skovhede
01c3b3f5c0
More work on a concurrent implementation of the backup process
2016-02-09 09:17:31 +01:00
Kenneth Skovhede