# ⚠️ Breaking
Many deprecated queue config options are removed (actually, they should
have been removed in 1.18/1.19).
If you see the fatal message when starting Gitea: "Please update your
app.ini to remove deprecated config options", please follow the error
messages to remove these options from your app.ini.
Example:
```
2023/05/06 19:39:22 [E] Removed queue option: `[indexer].ISSUE_INDEXER_QUEUE_TYPE`. Use new options in `[queue.issue_indexer]`
2023/05/06 19:39:22 [E] Removed queue option: `[indexer].UPDATE_BUFFER_LEN`. Use new options in `[queue.issue_indexer]`
2023/05/06 19:39:22 [F] Please update your app.ini to remove deprecated config options
```
Many options in `[queue]` are are dropped, including:
`WRAP_IF_NECESSARY`, `MAX_ATTEMPTS`, `TIMEOUT`, `WORKERS`,
`BLOCK_TIMEOUT`, `BOOST_TIMEOUT`, `BOOST_WORKERS`, they can be removed
from app.ini.
# The problem
The old queue package has some legacy problems:
* complexity: I doubt few people could tell how it works.
* maintainability: Too many channels and mutex/cond are mixed together,
too many different structs/interfaces depends each other.
* stability: due to the complexity & maintainability, sometimes there
are strange bugs and difficult to debug, and some code doesn't have test
(indeed some code is difficult to test because a lot of things are mixed
together).
* general applicability: although it is called "queue", its behavior is
not a well-known queue.
* scalability: it doesn't seem easy to make it work with a cluster
without breaking its behaviors.
It came from some very old code to "avoid breaking", however, its
technical debt is too heavy now. It's a good time to introduce a better
"queue" package.
# The new queue package
It keeps using old config and concept as much as possible.
* It only contains two major kinds of concepts:
* The "base queue": channel, levelqueue, redis
* They have the same abstraction, the same interface, and they are
tested by the same testing code.
* The "WokerPoolQueue", it uses the "base queue" to provide "worker
pool" function, calls the "handler" to process the data in the base
queue.
* The new code doesn't do "PushBack"
* Think about a queue with many workers, the "PushBack" can't guarantee
the order for re-queued unhandled items, so in new code it just does
"normal push"
* The new code doesn't do "pause/resume"
* The "pause/resume" was designed to handle some handler's failure: eg:
document indexer (elasticsearch) is down
* If a queue is paused for long time, either the producers blocks or the
new items are dropped.
* The new code doesn't do such "pause/resume" trick, it's not a common
queue's behavior and it doesn't help much.
* If there are unhandled items, the "push" function just blocks for a
few seconds and then re-queue them and retry.
* The new code doesn't do "worker booster"
* Gitea's queue's handlers are light functions, the cost is only the
go-routine, so it doesn't make sense to "boost" them.
* The new code only use "max worker number" to limit the concurrent
workers.
* The new "Push" never blocks forever
* Instead of creating more and more blocking goroutines, return an error
is more friendly to the server and to the end user.
There are more details in code comments: eg: the "Flush" problem, the
strange "code.index" hanging problem, the "immediate" queue problem.
Almost ready for review.
TODO:
* [x] add some necessary comments during review
* [x] add some more tests if necessary
* [x] update documents and config options
* [x] test max worker / active worker
* [x] re-run the CI tasks to see whether any test is flaky
* [x] improve the `handleOldLengthConfiguration` to provide more
friendly messages
* [x] fine tune default config values (eg: length?)
## Code coverage:
- This PR attempts to split our various DB tests into separate
pipelines.
- It splits up some of the extra feature-related tests rather than
having most of them in the MySQL test.
- It disables the race detector for some of the pipelines as well, as it
can cause slower runs and is mostly redundant when the pipelines just
swap DBs.
- It builds without SQLite support for any of the non-SQLite pipelines.
- It moves the e2e test to using SQLite rather than PG (partially
because I moved the minio tests to PG and that mucked up the test
config, and partially because it avoids another running service)
- It splits up the `go mod download` task in the Makefile from the tool
installation, as the tools are only needed in the compliance pipeline.
(Arguably even some of the tools aren't needed there, but that could be
a follow-up PR)
- SQLite is now the only arm64 pipeline, moving PG back to amd64 which
can leverage autoscaler
Should resolve#22010 - one thing that wasn't changed here but is
mentioned in that issue, unit tests are needed in the same pipeline as
an integration test in order to form a complete coverage report (at
least as far as I could tell), so for now it remains in a pipeline with
a DB integration test.
Please let me know if I've inadvertently changed something that was how
it was on purpose.
---
I will say sometimes it's hard to pin down the average time, as a
pipeline could be waiting for a runner for X minutes and that brings the
total up by X minutes as well, but overall this does seem to be faster
on average.
---------
Signed-off-by: jolheiser <john.olheiser@gmail.com>
Co-authored-by: techknowlogick <techknowlogick@gitea.io>
Fixes#12338
This allows use to talk to the API with our ssh certificate (and/or ssh-agent) without needing to fetch an API key or tokens.
It will just automatically work when users have added their ssh principal in gitea.
This needs client code in tea
Update: also support normal pubkeys
ref: https://tools.ietf.org/html/draft-cavage-http-signatures
Co-authored-by: Lunny Xiao <xiaolunwen@gmail.com>
Co-authored-by: zeripath <art27@cantab.net>
Signed-off-by: Andrew Thornton <art27@cantab.net>
* Ensure git tag tests and other create test repos in tmpdir
There are a few places where tests appear to reuse testing repos which
causes random CI failures.
This PR simply changes these tests to ensure that cloning always happens
into new temporary directories.
Fix#18444
* Change log root for integration tests to use the REPO_TEST_DIR
There is a potential race in the drone integration tests whereby test-mysql etc
will start writing to log files causing make test-check fail.
Fix#18077
Signed-off-by: Andrew Thornton <art27@cantab.net>
* Switch to non-deprecation setting
(Avoid by-default: "Deprecated fallback `[server]` `LFS_CONTENT_PATH` present. Use `[lfs]` `PATH` instead. This fallback will be removed in v1.18.0")
* Update all references
* Second attempt at preventing zombies
* Ensure that the pipes are closed in ssh.go
* Ensure that a cancellable context is passed up in cmd/* http requests
* Make cmd.fail return properly so defers are obeyed
* Ensure that something is sent to stdout in case of blocks here
Signed-off-by: Andrew Thornton <art27@cantab.net>
* placate lint
Signed-off-by: Andrew Thornton <art27@cantab.net>
* placate lint 2
Signed-off-by: Andrew Thornton <art27@cantab.net>
* placate lint 3
Signed-off-by: Andrew Thornton <art27@cantab.net>
* fixup
Signed-off-by: Andrew Thornton <art27@cantab.net>
* Apply suggestions from code review
Co-authored-by: 6543 <6543@obermui.de>
Co-authored-by: Lauris BH <lauris@nix.lv>
* Add register manual confirm settings option
The new settings option is used when manually approving new
registrations.
* Enable manual confirmation of new registered user
When manual registration confirmation is desired (by default `false`)
create new user in the database that is *not active*. The user must then
be activated manually.
This change speeds up the process of adding new confirmed users for
Gitea instances without external auth mechanism. (Currently the option
is to manually create new user by admin.)
* Update docs/content/doc/advanced/config-cheat-sheet.zh-cn.md
Co-authored-by: a1012112796 <1012112796@qq.com>
Co-authored-by: 6543 <6543@obermui.de>
Co-authored-by: a1012112796 <1012112796@qq.com>
Co-authored-by: Lunny Xiao <xiaolunwen@gmail.com>
* Fix test
* Add no queue for test only
* improve code
* Auto watch whatever branch operation
* Fix lint
* Rename noqueue to immediate
* Remove old PushUpdate function
* Fix tests
Co-authored-by: zeripath <art27@cantab.net>
Co-authored-by: techknowlogick <techknowlogick@gitea.io>
Some integration test data is being put outside of the integration test folders which means it isn't deleted properly with make clean. Update config so it works as expected.
Co-authored-by: John Olheiser <john.olheiser@gmail.com>
Co-authored-by: techknowlogick <techknowlogick@gitea.io>
* Add option to use /tmp for test repositories
* Fix exit status
* Add feedback about using tmp repos
Co-authored-by: Guillermo Prandi <guillep2k@users.noreply.github.com>
Co-authored-by: Lunny Xiao <xiaolunwen@gmail.com>
* Add support for database schema
* Require setting search_path for the db user
* Add schema setting to admin/config.tmpl
* Use a schema different from default for psql tests
* Update postgres scripts to use custom schema
* Update to xorm/core 0.7.3 and xorm/xorm c37aff9b3a
* Fix migration test
Co-authored-by: Antoine GIRARD <sapk@users.noreply.github.com>
Co-authored-by: Lunny Xiao <xiaolunwen@gmail.com>
This PR fixes#7598 by providing a configurable way of signing commits across the Gitea instance. Per repository configurability and import/generation of trusted secure keys is not provided by this PR - from a security PoV that's probably impossible to do properly. Similarly web-signing, that is asking the user to sign something, is not implemented - this could be done at a later stage however.
## Features
- [x] If commit.gpgsign is set in .gitconfig sign commits and files created through repofiles. (merges should already have been signed.)
- [x] Verify commits signed with the default gpg as valid
- [x] Signer, Committer and Author can all be different
- [x] Allow signer to be arbitrarily different - We still require the key to have an activated email on Gitea. A more complete implementation would be to use a keyserver and mark external-or-unactivated with an "unknown" trust level icon.
- [x] Add a signing-key.gpg endpoint to get the default gpg pub key if available
- Rather than add a fake web-flow user I've added this as an endpoint on /api/v1/signing-key.gpg
- [x] Try to match the default key with a user on gitea - this is done at verification time
- [x] Make things configurable?
- app.ini configuration done
- [x] when checking commits are signed need to check if they're actually verifiable too
- [x] Add documentation
I have decided that adjusting the docker to create a default gpg key is not the correct thing to do and therefore have not implemented this.
resolves git conflicts from #3896 (credit to @belak, in case github doesn't keep original author during squash)
Co-Authored-By: Matti Ranta <techknowlogick@gitea.io>
* Rename BaseLogger to WriterLogger to help the creation of other providers
* Don't export ColorBytes and ResetBytes from ColoredValues
* Make integration tests only print logs if they fail
* check can color before coloring
* I always forget about MSSQL
* Oh and use LEVEL in sqlite.ini
* Make the test logger log at info - as it means you see the router
* Remove empty expected changes
* Make the migrations quieter too
* Don't display SQL on error - it can be looked at in the file logs if necessary
* Fix skip when using onGiteaRun
* Create log mailer for testing email settings
Signed-off-by: Andrew Thornton <art27@cantab.net>
* Switch on the log mailer for the integration tests
This ensures that the sending mail process works
Signed-off-by: Andrew Thornton <art27@cantab.net>
* rename the from user for mysql/mssql
* rename log sender to dummy sender
* update the integration tests
* test: integration add git cli tests
Extracted form for easing review process and debug #3152
* test: integration add git cli big file commit
* fix: Don't rewrite key if internal server
* Implement routes
* move to api/sdk and create model
* Implement add + list
* List return 200 empty list no 404
* Add verify lfs lock api
* Add delete and start implementing auth control
* Revert to code.gitea.io/sdk/gitea vendor
* Apply needed check for all lfs locks route
* Add simple tests
* fix lint
* Improve tests
* Add delete test + fix
* Add lfs ascii header
* Various fixes from review + remove useless code + add more corner case testing
* Remove repo link since only id is needed.
Save a little of memory and cpu time.
* Improve tests
* Use TEXT column format for path + test
* fix mispell
* Use NewRequestWithJSON for POST tests
* Clean path
* Improve DB format
* Revert uniquess repoid+path
* (Re)-setup uniqueness + max path length
* Fixed TEXT in place of VARCHAR
* Settle back to maximum VARCHAR(3072)
* Let place for repoid in key
* Let place for repoid in key
* Let place for repoid in key
* Revert back
Avoids override of source file upon running `make test-pgsql`
Long story: settings.NewContext *adds* an INTERNAL_TOKEN if none
is found; the other ini files under integration/ all have an
INTERNAL_TOKEN
FuXiaoHei
wxiaoguang
John Olheiser
Kyle D
Wim
Lunny Xiao
zeripath
Gusted
KN4CK3R
Jiri Vlasak
techknowlogick
mrsdizzie
guillep2k
Antoine GIRARD
Lauris BH
Ethan Koenig
Thomas Boerger
Sandro Santilli
Ethan Koenig
