Summary:
Ref T12932. For long-lived installs, one of the largest tables tends to be the hunk data table. Although it doesn't grow tremendously fast, it's also well suited to storage in Files instead of the database (infrequent access, relatively large blobs of data, mostly one-at-a-time access), and earlier work anticipated eventually adding support for Files storage.
Make Files storage work, and provide `bin/differential migrate-hunk` to manually test/migrate hunks. This is currently the only way hunks get moved to file storage, but I expect to add a GC step which moves them to File storage after 30 days shortly.
The immediate motivation for this is to relieve storage pressure on db001/db002 so we have more headroom for deploying the Ferret engine and its larger indexes (see also T12819).
Test Plan:
- Used `bin/differential migrate-hunk` to move a hunk to and from file storage, verified it survived intact.
- Downloaded the actual stored file, sanity-checked it. Verified permissions.
- Destroyed a diff with `bin/remove destroy`, saw the hunk and file storage destroyed.
- Verified that going from file -> text destroys the old file properly with `migrate-hunk --trace ...`.
Reviewers: chad
Reviewed By: chad
Maniphest Tasks: T12932
Differential Revision: https://secure.phabricator.com/D18584
Summary: Ref T8475. This gets rid of most of the old "legacy hunk" code. I'll nuke the rest (and drop the old table) once we're more sure that we're in the clear.
Test Plan: Browsed Differential.
Reviewers: chad
Reviewed By: chad
Maniphest Tasks: T8475
Differential Revision: https://secure.phabricator.com/D17040
Summary:
Ref T10537. For Nuance, I want to introduce new sources (like "GitHub" or "GitHub via Nuance" or something) but this needs to modularize eventually.
Split ContentSource apart so applications can add new content sources.
Test Plan:
This change has huge surface area, so I'll hold it until post-release. I think it's fairly safe (and if it does break anything, the breaks should be fatals, not anything subtle or difficult to fix), there's just no reason not to hold it for a few hours.
- Viewed new module page.
- Grepped for all removed functions/constants.
- Viewed some transactions.
- Hovered over timestamps to get content source details.
- Added a comment via Conduit.
- Added a comment via web.
- Ran `bin/storage upgrade --namespace XXXXX --no-quickstart -f` to re-run all historic migrations.
- Generated some objects with `bin/lipsum`.
- Ran a bulk job on some tasks.
- Ran unit tests.
{F1190182}
Reviewers: chad
Reviewed By: chad
Maniphest Tasks: T10537
Differential Revision: https://secure.phabricator.com/D15521
Summary:
Ref T9319. See D14967. As before, this is making a deeply-buried, complex operation easier to test by providing a CLI command.
This adds `bin/differential attach-commit rXnnnn Dnnnn` to pretend that `rXnnnn` was just committed and matched `Dnnnn`.
Test Plan:
- Ran `bin/differential attach-commit X Y` for several different values, saw updates in the UI.
- Faked the message parser to make sure stuff still worked there.
Reviewers: chad
Reviewed By: chad
Maniphest Tasks: T9319
Differential Revision: https://secure.phabricator.com/D14968
Summary:
Ref T9319. When we discover a commit, we sometimes update the corresponding revision with a "this is the actual committed change" diff and send out a link to the changes between review and commit.
This is currently very difficult to test, because it only happens the first time and you have to either go set up a bunch of objects or add a bunch of special casing to the parser to hit the workflow.
I'm making some changes to how it pulls file content. To make those changes easier to test, first start extracting this stuff so the code can be run with `bin/differential extract ...` instead of needing to do a bunch of more complicated setup steps.
Test Plan:
- Ran `bin/differential extract ...` to extract diffs from commits.
- Forced my way through the daemon workflow by faking out a bunch of flags, got a clean extract + attach + update. After this patch, this should rarely be necessary.
Reviewers: chad
Reviewed By: chad
Maniphest Tasks: T9319
Differential Revision: https://secure.phabricator.com/D14967
Summary: This `break` statement causes `./bin/hunks migrate` to only migrate one-hunk-at-a-time, which is unnecessary and slightly misleading. Instead, allow the script to migrate //all// legacy hunks to modern storage. In particular, this means that we can recommend that installs run this command sometime before D13222 is landed.
Test Plan: It's a pain to setup the data necessary to test this, but this is identical to the change that I made on our production install when I migrated our hunk storage.
Reviewers: epriestley, #blessed_reviewers
Reviewed By: epriestley, #blessed_reviewers
Subscribers: epriestley, Korvin
Differential Revision: https://secure.phabricator.com/D13288
Summary: Ref T4045. Ref T5179. When saving a modern hunk, deflate it if we have the function and deflating it will save a nontrivial number of bytes.
Test Plan:
- Used `bin/hunks migrate` to move some hunks over, saw ~70-80% compression on most standard hunks.
- Viewed changesets using compressed hunks.
- Profiled `gzinflate()` and verified the cost is trivial (<< 1ms) at least for normal diffs.
Reviewers: btrahan
Reviewed By: btrahan
Subscribers: epriestley
Maniphest Tasks: T4045, T5179
Differential Revision: https://secure.phabricator.com/D9292
Summary:
Ref T4045. Ref T5179. While we'll eventually need to force a migration, we can let installs (particularly large installs) do an online migration for now. This moves hunks to the new storage format one at a time.
(Note that nothing writes to the new store yet, so this is the only way to populate it.)
WARNING: Installs, don't run this yet! It won't compress the data. Wait until it can also do compression.
Test Plan: Added a `break;` after migrating one row and moved a few rows over. Spot checked them in the database and viewed the affected diffs.
Reviewers: btrahan
Reviewed By: btrahan
Subscribers: epriestley
Maniphest Tasks: T4045, T5179
Differential Revision: https://secure.phabricator.com/D9291
