dillard
343389b98b
Changelog: the second read covers the database phase too
2026-10-22 19:48:22 +02:00
dillard
133f5f9d34
Schedule: say why a server zone of UTC is passed over
2026-10-22 19:47:30 +02:00
dillard
cf4bd9db7f
Review fixes: the database restore also reads a damaged blob once more; schedule conversion right on days the clocks change; the field follows a changed schedule time zone
2026-10-22 19:45:49 +02:00
dillard
67cc350b8c
Daily backup time in the administrator's time zone ( #19 ); a time saved before is converted once so it shows as it ran
2026-10-22 16:38:54 +02:00
dillard
fb5b234e8a
Restore: a blob that does not decrypt as first read is read once more on a request of its own before the restore gives up
2026-10-22 14:43:48 +02:00
dillard
fd12c586b6
Roadmap: restore mode for hosted Nextcloud, read-ahead for S3 and SMB; changelog for rc6
2026-10-22 13:07:09 +02:00
dillard
6ffe979ea7
Merge branch 'polish' into rc6
2026-10-22 13:06:59 +02:00
dillard
a03b7354b3
README: tested disaster recoveries, backend measurements, restores without shell access; guide lists shell access; changelog for rc6
2026-10-22 13:06:49 +02:00
dillard
be1c628f07
Full-restore test: install Mail and Deck with up to three tries (the app store is slow at times)
2026-10-22 03:38:30 +02:00
dillard
a8355df12e
Full restore: tables the restore creates get this installation's charset and collation (utf8mb4 on MySQL/MariaDB); test with an emoji
2026-10-22 03:36:46 +02:00
f0537f0f2c
Merge pull request #18 from dillardblom/rc5
...
CI / unit (8.2) (push) Has been cancelled
CI / unit (8.3) (push) Has been cancelled
CI / unit (8.4) (push) Has been cancelled
CI / psalm (push) Has been cancelled
CI / integration (mariadb, 32) (push) Has been cancelled
CI / integration (mariadb, 33) (push) Has been cancelled
CI / integration (mariadb, 34) (push) Has been cancelled
CI / integration (mariadb, 35) (push) Has been cancelled
CI / integration (pgsql, 32) (push) Has been cancelled
CI / integration (pgsql, 33) (push) Has been cancelled
CI / integration (pgsql, 34) (push) Has been cancelled
CI / integration (pgsql, 35) (push) Has been cancelled
CI / integration (sqlite, 32) (push) Has been cancelled
CI / integration (sqlite, 33) (push) Has been cancelled
CI / integration (sqlite, 34) (push) Has been cancelled
CI / integration (sqlite, 35) (push) Has been cancelled
CI / integration-targets (push) Has been cancelled
CI / integration-user (32) (push) Has been cancelled
CI / integration-user (33) (push) Has been cancelled
CI / integration-user (34) (push) Has been cancelled
CI / integration-user (35) (push) Has been cancelled
CI / integration-full-restore (push) Has been cancelled
Release 1.0.0-rc.5
2026-10-22 02:28:05 +02:00
dillard
fd0f786f5f
Full restore: a run again after an interrupted restore keeps maintenance mode on when its check refuses; retry text about apps
2026-10-22 01:59:01 +02:00
dillard
2eabdf44d5
README and store description: background jobs must run on system cron, with how to set it up; rc5 date
2026-10-22 01:03:52 +02:00
dillard
61e2a28b67
Full restore: maintenance mode on before the checks (off again when they refuse); the printed retry command is complete
2026-10-22 00:53:08 +02:00
dillard
7ed53ad0eb
Full restore: after an error maintenance mode stays on, with the one line to switch it off and run again; maintenance that was on before stays on
2026-10-22 00:36:58 +02:00
dillard
bdcb4f0cae
Full restore runs in maintenance mode, switched off again at the end (also after an error)
2026-10-22 00:31:34 +02:00
dillard
2616b6e543
Init guard also counts snapshots in the trash and the catalog as backup data
2026-10-21 22:42:43 +02:00
dillard
9bc4e4a4ab
Never create a repository where backup data lies but no config could be read
2026-10-21 22:25:48 +02:00
dillard
827e46381e
Log a snapshot that cannot be adopted when a location is added; full-restore test prints what target:add and backup:list said when the snapshot is missing
2026-10-21 22:08:14 +02:00
dillard
8eb6c4c033
Index: check known index files spread over the list
2026-10-21 20:56:45 +02:00
dillard
d623416306
Index: an empty listing is taken only when known index files are gone too
2026-10-21 20:49:32 +02:00
dillard
cd251cffb3
PLAN: SMB index reads one at a time (1.1.0)
2026-10-21 20:02:43 +02:00
dillard
0b4872599d
PLAN: the 1.0.1 cron PR closes #13
2026-10-21 19:51:57 +02:00
dillard
95184dadc7
Index: an empty listing is checked (config present, listed once more) before it is taken, so a repository another server emptied is accepted and a hiccup recovers
2026-10-21 19:51:46 +02:00
dillard
9b6bd7d2ab
Release 1.0.0-rc.5 (app version 0.9.11)
2026-10-21 13:10:03 +02:00
dillard
2fcd38f478
Full-restore test: Mail and Deck checks only when the origin had them (not on a SQLite origin)
2026-10-21 13:09:54 +02:00
dillard
b7d09c9784
PLAN: 1.1.0 faster SMB (local and long-line measurements)
2026-10-21 09:27:15 +02:00
dillard
177361acae
Index load keeps the index when it refuses an empty listing; counters of unchanged tables are repaired too, never moved backwards
2026-10-21 09:08:50 +02:00
dillard
24686897c3
PLAN: DR tests across databases (integration for every crossing; S3 sim onto MariaDB, SMB sim onto SQLite before 1.0.0; nc01 onto MariaDB after)
2026-10-20 23:12:19 +02:00
dillard
85ea0acc50
Full-restore test: restore onto another database (to-pgsql, to-mysql, to-sqlite), every crossing in CI
2026-10-20 23:10:25 +02:00
dillard
ac57e31c93
Full restore onto PostgreSQL: counters of restored tables go after their data when the backup records none (MySQL, SQLite); test MariaDB onto PostgreSQL
2026-10-20 23:04:53 +02:00
dillard
315e83b847
PLAN: 1.0.1 progress in the database phase and a waiting notice, so a backup does not look stuck
2026-10-20 20:49:44 +02:00
dillard
e461c48056
CI: back to Nextcloud 35 for the full-restore test (the failures were a slow app store, not the version)
2026-10-19 22:18:33 +02:00
dillard
d504067cd8
Index: an empty listing is an error in checks and cleanups too; refresh() fetches only index files it has not loaded; full-restore CI on Nextcloud 35.0.0
2026-10-19 22:13:26 +02:00
dillard
c37b26ea38
Full restore: an empty listing never empties the index; read a failed pack once more; recognise and replace SFTP connections a Storage Box dropped quietly; version 0.9.11
2026-10-19 21:53:05 +02:00
dillard
b728db36c7
README: roadmap
2026-10-19 21:35:23 +02:00
dillard
6ed0a555cf
PLAN: label field next to Back up now (1.0.1); per-location schedules, every N hours, locations one after the other, editable notes (1.1.0)
2026-10-19 21:26:20 +02:00
dillard
221ed9dc17
PLAN: the cron warning says cron is needed for all backups, links to how to set it up, and that without it a backup started with the button only moves while the page is open
2026-10-19 21:18:23 +02:00
dillard
285153a968
PLAN: 1.0.1 cron warning says to keep the page open under AJAX, with a link to the Nextcloud docs
2026-10-19 21:16:23 +02:00
dillard
6c8a78f6bb
PLAN: 1.0.1 background jobs without system cron: warn, and let the open admin page move a backup
2026-10-19 20:41:11 +02:00
0accd5a3f2
Merge pull request #16 from dillardblom/rc4
...
CI / unit (8.2) (push) Has been cancelled
CI / unit (8.3) (push) Has been cancelled
CI / unit (8.4) (push) Has been cancelled
CI / psalm (push) Has been cancelled
CI / integration (mariadb, 32) (push) Has been cancelled
CI / integration (mariadb, 33) (push) Has been cancelled
CI / integration (mariadb, 34) (push) Has been cancelled
CI / integration (mariadb, 35) (push) Has been cancelled
CI / integration (pgsql, 32) (push) Has been cancelled
CI / integration (pgsql, 33) (push) Has been cancelled
CI / integration (pgsql, 34) (push) Has been cancelled
CI / integration (pgsql, 35) (push) Has been cancelled
CI / integration (sqlite, 32) (push) Has been cancelled
CI / integration (sqlite, 33) (push) Has been cancelled
CI / integration (sqlite, 34) (push) Has been cancelled
CI / integration (sqlite, 35) (push) Has been cancelled
CI / integration-targets (push) Has been cancelled
CI / integration-user (32) (push) Has been cancelled
CI / integration-user (33) (push) Has been cancelled
CI / integration-user (34) (push) Has been cancelled
CI / integration-user (35) (push) Has been cancelled
CI / integration-full-restore (push) Has been cancelled
Release 1.0.0-rc.4
2026-10-19 11:04:15 +02:00
dillard
583592f69b
Release 1.0.0-rc.4 (app version 0.9.10)
2026-10-19 10:34:01 +02:00
dillard
a384ab41bb
App store description: Ensembia line
2026-10-19 10:31:21 +02:00
dillard
9b85221e50
Read ahead: a plan that cannot be made leaves the restore to the usual reads; only a whole buffered SFTP packet counts as an answer waiting
2026-10-19 08:54:13 +02:00
dillard
c95f552cc6
Read ahead over SFTP: one connection with 32 requests in flight by default, more with restore_sftp_connections; a Storage Box gets slower with parallel readers
2026-10-19 07:46:12 +02:00
dillard
6121f684e9
Full restore: say to run occ upgrade before rerunning a restore that stopped after the database; remove empty spool folders of a stopped restore
2026-10-19 05:34:54 +02:00
dillard
43a71684f9
Read ahead over SFTP: take answers phpseclib has read already instead of waiting up to 10 s on the socket
2026-10-19 05:28:26 +02:00
dillard
88b719879d
Release candidate 4 (app version 0.9.10): full restore over SFTP reads ahead; SFTP leg in the full-restore test
2026-10-19 03:51:21 +02:00
dillard
1d7f7e6423
Read ahead: resolve paths before opening connections; a failing readAhead() leaves no spool folder
2026-10-19 01:52:58 +02:00
dillard
ccbbf01c54
Full restore over SFTP: read packs ahead on four connections at once
...
A restore read one pack at a time over one SFTP connection, which over a long
line is bound by the round trip and by what one TCP connection gets: about
2-3 MiB/s from a Hetzner Storage Box to Canada, with PHP mostly waiting.
Repository::restore() now first lists the ranges of packs the restore needs,
in order (ReadAheadPlan), and the location reads them ahead into unlinked
files in a spool folder in the target (IReadsAhead). Over SFTP,
SftpRangeFetcher does that on four connections of its own, 16 requests in
flight each, taking the answers of whichever connection has one. A range that
fails, a damaged copy, or a blob that moved since the plan is read the usual
way, which reports what is wrong. Other locations restore as before.
On a test line with 120 ms RTT and 0.2% loss, a 1.6 GiB restore went from
5.1 to 17.8 MiB/s, all files identical.
2026-10-19 01:40:49 +02:00
dillard
Dillard Blom