Commit Graph

151 Commits

Author SHA1 Message Date
Gerhard Schlager 58c218a4bf
FIX: Remap old S3 endpoints during backup restore (#12276)
It also starts outputting exceptions on the console.
2021-03-03 21:10:09 +01:00
Gerhard Schlager 154bfcf750
FEATURE: Include details about S3 backup storage errors (#12257) 2021-03-02 15:29:37 +01:00
Gerhard Schlager 4b05fc2d2d
FIX: Restoring backup via UI was broken (#12089) 2021-02-15 16:46:44 +01:00
Gerhard Schlager 0b05302cfe FIX: Restoring could fail due to missing path 2021-02-09 17:28:03 +01:00
Gerhard Schlager 4f5ea4fbde FIX: Restoring backup could fail due to missing uploads
Clearing theme and emoji cache might require uploaded files.
2021-02-09 17:28:03 +01:00
Gerhard Schlager 4d719725c8
FEATURE: Allow overriding the backup location when restoring via CLI (#12015)
You can use `discourse restore --location=local FILENAME` if you want to restore a backup that is stored locally even though the `backup_location` has the value `s3`.
2021-02-09 16:02:44 +01:00
Gerhard Schlager d5ef6188ed
PERF: Disable Sidekiq only during database restore (#10857)
It pauses Sidekiq, clears Redis (namespaced to the current site), clears Sidekiq jobs for the current site, restores the database and unpauses Sidekiq. Previously it stayed paused until the end of the restore.

Redis is cleared because we don't want any old data lying around (e.g. old Sidekiq jobs). Most data in Redis is prefixed with the name of the multisite, but Sidekiq jobs in a multisite are all stored in the same keys. So, deleting those jobs requires a little bit more logic.
2020-10-16 15:19:02 +02:00
Gerhard Schlager 1febf11362 FIX: Backup didn't work anymore after a running backup was canceled 2020-10-13 19:48:53 +02:00
Gerhard Schlager ac70c48be4 FIX: Prevent "uploads are missing in S3" alerts after restoring a backup
After restoring a backup it takes up to 48 hours for uploads stored on S3 to appear in the S3 inventory. This change prevents alerts about missing uploads by preventing the EnsureS3UploadsExistence job from running in the first 48 hours after a restore. During the restore it  deletes the count of missing uploads from the PluginStore, so that an alert isn't triggered by an old number.
2020-09-10 21:37:48 +02:00
Gerhard Schlager c0cd69d280
DEV: Restoring backups didn't work on macOS
This is a follow-up for f51ccea028 because specs were failing on macOS. macOS uses bsdtar by default and the --transform option isn't supported by bsdtar. But the -s option is and it works the same.
2020-08-27 19:08:47 +02:00
Gerhard Schlager f51ccea028
FIX: Backups should use relative paths for local uploads
This also ensures that restoring a backup works when it was created with the wrong upload paths in the time between ab4c0a4970 (shortly after v2.6.0.beta1) and this fix.
2020-08-21 15:22:28 +02:00
Guo Xiang Tan c7f1777ac9
DEV: No need to pause Sidekiq during backups.
This was done for historical reasons where we stop all writes going to
the DB during a backup by setting the database to readonly mode.
2020-07-29 10:52:31 +08:00
Gerhard Schlager ab4c0a4970 FEATURE: Create SQL-only backup if there are no uploads
It doesn't make sense to compress the database dump twice if the backup doesn't contain any uploaded files.
2020-07-07 16:23:47 +02:00
Gerhard Schlager fc8e842773 FIX: Sometimes not all output of psql was logged during restores
There was a race condition which could prevent Discourse from logging the last couple of lines of output from psql.
2020-06-30 16:52:50 +02:00
Gerhard Schlager 859d9b75a7 FIX: Restoring backup from PG12 could fail on PG10
The `EXECUTE FUNCTION` syntax for `CREATE TRIGGER` statements was introduced in PostgreSQL 11. We need to replace `EXECUTE FUNCTION` with `EXECUTE PROCEDURE` in order to be able to restore backups created with PG12 on PG10.
2020-06-16 16:04:14 +02:00
Andrew Schleifer b2c94cc8ea FIX: do not migrate backups in the new prefix 2020-06-12 02:56:07 +00:00
Andrew Schleifer 74d28a43d1
new S3 backup layout (#9830)
* DEV: new S3 backup layout

Currently, with $S3_BACKUP_BUCKET of "bucket/backups", multisite backups
end up in "bucket/backups/backups/dbname/" and single-site will be in
"bucket/backups/".

Both _should_ be in "bucket/backups/dbname/"

- remove MULTISITE_PREFIX,
- always include dbname,
- method to move to the new prefix
- job to call the method

* SPEC: add tests for `VacateLegacyPrefixBackups` onceoff job.

Co-authored-by: Vinoth Kannan <vinothkannan@vinkas.com>
2020-05-29 00:28:23 +05:30
Michael Brown d9a02d1336
Revert "Revert "Merge branch 'master' of https://github.com/discourse/discourse""
This reverts commit 20780a1eee.

* SECURITY: re-adds accidentally reverted commit:
  03d26cd6: ensure embed_url contains valid http(s) uri
* when the merge commit e62a85cf was reverted, git chose the 2660c2e2 parent to land on
  instead of the 03d26cd6 parent (which contains security fixes)
2020-05-23 00:56:13 -04:00
Jeff Atwood 20780a1eee Revert "Merge branch 'master' of https://github.com/discourse/discourse"
This reverts commit e62a85cf6f, reversing
changes made to 2660c2e21d.
2020-05-22 20:25:56 -07:00
Gerhard Schlager 0a700d81fc FIX: Restoring backups could fail for database dumps > 8GiB
This is a temporary fix until we ship a new image with bsdtar.
2020-05-19 22:36:59 +02:00
Gerhard Schlager 6d5e9db883 FIX: Restoring backup didn't clear cached translation overrides 2020-05-18 18:51:51 +02:00
Dan Ungureanu 3d9c320aab
PERF: Cache Category.subcategory_ids (#9350)
Also reset category cache after backup restore.
2020-04-09 15:42:24 +03:00
Gerhard Schlager ad6709772a PERF: Backup with lots of uploads stored on S3 was slow
Creating the backup needs a lot more disk space after this change, but it is a lot faster.
2020-04-03 18:13:34 +02:00
Gerhard Schlager 13b4eb9cce FIX: Restore failed if schema contained objects not owned by the current DB user 2020-04-01 18:04:43 +02:00
Roman Rizzi 57321b90f0
Completely remove read only mode during backups (#9279) 2020-03-27 07:38:55 -03:00
Gerhard Schlager 8022e51179 FIX: Failed to restore backups from versions without translation overrides
Rails calls I18n.translate during initialization and by default translation overrides are used. Database migrations would fail if the system tried to migrate from an old version that didn't have the `translation_overrides` table with all its columns yet.

This makes restoring really old backups work again. Running `DISABLE_TRANSLATION_OVERRIDES=1 rake db:migrate` will allow you to upgrade such an old database as well.
2020-03-14 00:00:22 +01:00
Roman Rizzi 87687c0819
Drop unnecessary readonly_during_backup setting (#9112) 2020-03-06 14:29:00 -03:00
Gerhard Schlager 8fa8bab9ff FIX: Don't optimize icons during db:migrate when restoring backup
Uploads are extracted after the DB migration, so this could lead to a failure during the restore. Site icons get optimized after extracting uploads.
2020-03-04 16:59:49 +01:00
Gerhard Schlager 7c30986b5e FIX: Failed to notify user after restoring backup 2020-01-25 22:07:41 +01:00
Gerhard Schlager f216c6d60b FEATURE: Drop "backup" schema 7 days after restore
The "backup" schema is used to rollback a failed restore. It isn't useful after a longer period of time and turns into a waste of disk space.
2020-01-16 17:48:47 +01:00
Gerhard Schlager 5e3fc31f2c DEV: Less hacky way of rolling back DB changes
Some specs use psql to test database restores and dropping the table after the test needs to happen outside of rspec because of transactions. The previous attempt lead to some changes to be stored in the test database.
2020-01-15 23:37:42 +01:00
Gerhard Schlager 68a7ae3091 REFACTOR: Simplify backup version check
Adds specs for inalid version number in metadata file.
Follow-up to c3cd2389fe
2020-01-15 23:37:40 +01:00
Martin Brennan c3cd2389fe SECURITY: use strict JSON parsing when parsing backup metadata 2020-01-15 11:24:41 +01:00
Gerhard Schlager e474cda321 REFACTOR: Restoring of backups and migration of uploads to S3 2020-01-14 11:41:35 +01:00
Vinoth Kannan 3b7f5db5ba
FIX: parallel spec system needs a dedicated upload folder for each worker. (#8547) 2019-12-18 11:21:57 +05:30
David Taylor 481efebe76
DEV: Update backup/restore pipeline to avoid `cd` (#8347) 2019-11-13 15:52:28 +00:00
Justin DiRose c3f06943c7
FIX: Account for empty uploads directory upon backup restore (#8262)
This commit fixes a case where backup restores would fail if the uploads/default directory is empty.
2019-10-30 09:33:07 -05:00
Krzysztof Kotlarek f530378df3 FIX: Restore for non-multisite is not raising an error on reconnect step (#8237)
That commit introduced a bug to the system: f69dacf979

Restore works fine for multisite, however, stopped working for non-multisite.

Reason for that was that `establish_connection` method got a check if the multisite instance is available:
```
    def self.instance
      @instance
    end

    def self.establish_connection(opts)
      @instance.establish_connection(opts) if @instance
    end
```
However, the reload method don't have that check
```
    def self.reload
      @instance = new(instance.config_filename)
    end
```

To solve it, let's ensure we are in a multisite environment before call reload
2019-10-24 11:46:22 +11:00
Krzysztof Kotlarek f69dacf979 FIX: Reconnect in restore process connects to correct DB (#8218)
Simplified flow of restore is like that
```
migrate_database
reconnect
extract_uploads
```

Problem with incorrect current database started with this fix https://github.com/discourse/discourse/commit/025d4ee91f4

Dump task is reconnecting to default database https://github.com/rails/rails/blob/master/activerecord/lib/active_record/railties/databases.rake#L429

And then, we are trying to reconnect to the original database with that code:
```
def reconnect_database
  log "Reconnecting to the database..."
  RailsMultisite::ConnectionManagement::establish_connection(db: @current_db)
end
```

This reconnect is not switching us back to correct database because of that check
https://github.com/discourse/rails_multisite/blob/master/lib/rails_multisite/connection_management.rb#L181
Basically, it finds existing handler and it thinks that we are connected to correct DB and this step can be skipped.

To solve it, we can reload RailsMultisite::ConnectionManagement which creates a new instance of that class
https://github.com/discourse/rails_multisite/blob/master/lib/rails_multisite/connection_management.rb#L38
2019-10-23 17:23:50 +11:00
Roman Rizzi 01bc465db8
DEV: Split max decompressed setting for themes and backups (#8179) 2019-10-11 14:38:10 -03:00
Roman Rizzi 5357ab3324
SECURITY: Safely decompress backups when restoring. (#8166)
* SECURITY: Safely decompress backups when restoring.

* Fix tests and update theme_controller_spec to work with zip files instead of .tar.gz
2019-10-09 11:41:16 -03:00
Krzysztof Kotlarek 427d54b2b0 DEV: Upgrading Discourse to Zeitwerk (#8098)
Zeitwerk simplifies working with dependencies in dev and makes it easier reloading class chains. 

We no longer need to use Rails "require_dependency" anywhere and instead can just use standard 
Ruby patterns to require files.

This is a far reaching change and we expect some followups here.
2019-10-02 14:01:53 +10:00
Gerhard Schlager 1f118b1309 FEATURE: Allow plugins to manipulate site settings during backup restore 2019-08-22 22:41:26 +02:00
Gerhard Schlager d686318133 FIX: Prevent failed remaps during restores
Additional changes:
* Verbose logging of remaps during restores
* Exclude the backup_metadata table from restores
2019-08-12 17:15:01 +02:00
Gerhard Schlager 7cb51d0e40 FIX: Create readonly functions during backup
Temporarily recreate already dropped functions in the discourse_functions schema in order to allow restoring of backups which still reference dropped functions.
2019-08-09 11:39:46 +02:00
Michael Brown 31f583855a DEV: pull static check out of loop
* followup to 08b28680
* as per https://review.discourse.org/t/4713/2
2019-07-23 17:18:16 -04:00
Gerhard Schlager 68b082e1a4 FIX: Ensure that jobs don't run immediately after migrate_to_s3 2019-07-23 17:42:12 +02:00
Gerhard Schlager b73bd7fc1b FIX: Always backup local uploads in addition to files stored on S3 2019-07-19 15:13:05 +02:00
Blake Erickson b0c92bb0b9 REFACTOR: Clean up parameterized title
Follow up to [FIX: Empty backup names with unicode site titles][1]

- Use .presence - "It's cleaner"
- Update spec to use System.system_user so it is more readable

[1]: c8661674d4
2019-07-18 15:49:16 -06:00
Blake Erickson c8661674d4 FIX: Empty backup names with unicode site titles
If a site title contains unicode it may end up with an empty backup
filename because of the rails `parameterize` method we are calling.

This fix ensures that the backup filenames default to "discourse" if the
parameterized site title is empty.

Bug reported [here][1].

[1]: https://meta.discourse.org/t/backup-checksum-and-backup-name-missing-when-unicode-site-name/123192?u=blake
2019-07-17 17:07:10 -06:00