Backport #28942 by @Zettat123
In #28691, schedule plans will be deleted when a repo's actions unit is
disabled. But when the unit is enabled, the schedule plans won't be
created again.
This PR fixes the bug. The schedule plans will be created again when the
actions unit is re-enabled
---------
Co-authored-by: Zettat123 <zettat123@gmail.com>
Co-authored-by: Lunny Xiao <xiaolunwen@gmail.com>
Backport #28886 by @KN4CK3R
The method can't be called with an outer transaction because if the user
is not a collaborator the outer transaction will be rolled back even if
the inner transaction uses the no-error path.
`has == 0` leads to `return nil` which cancels the transaction. A
standalone call of this method does nothing but if used with an outer
transaction, that will be canceled.
Co-authored-by: KN4CK3R <admin@oldschoolhack.me>
Fix reverting a merge commit failing (#28794) (#28825)
Backport https://github.com/go-gitea/gitea/pull/28794
Fixes #22236
---
Error occurring currently while trying to revert commit using read-tree
-m approach:
> 2022/12/26 16:04:43 ...rvices/pull/patch.go:240:AttemptThreeWayMerge()
[E] [63a9c61a] Unable to run read-tree -m! Error: exit status 128 -
fatal: this operation must be run in a work tree
> - fatal: this operation must be run in a work tree
We need to clone a non-bare repository for `git read-tree -m` to work.
bb371aee6e
adds support to create a non-bare cloned temporary upload repository.
After cloning a non-bare temporary upload repository, we [set default
index](https://github.com/go-gitea/gitea/blob/main/services/repository/files/cherry_pick.go#L37)
(`git read-tree HEAD`).
This operation ends up resetting the git index file (see investigation
details below), due to which, we need to call `git update-index
--refresh` afterward.
Here's the diff of the index file before and after we execute
SetDefaultIndex: https://www.diffchecker.com/hyOP3eJy/
Notice the **ctime**, **mtime** are set to 0 after SetDefaultIndex.
You can reproduce the same behavior using these steps:
```bash
$ git clone https://try.gitea.io/me-heer/test.git -s -b main
$ cd test
$ git read-tree HEAD
$ git read-tree -m 1f085d7ed81f085d7ed89933caed00
error: Entry '1' not uptodate. Cannot merge.
```
After which, we can fix like this:
```bash
$ git update-index --refresh
$ git read-tree -m 1f085d7ed81f085d7ed89933caed00
```
Fix #28157
Backport #28691
This PR fix the possible bugs about actions schedule.
- Move `UpdateRepositoryUnit` and `SetRepoDefaultBranch` from models to
service layer
- Remove schedules plan from database and cancel waiting & running
schedules tasks in this repository when actions unit has been disabled
or global disabled.
- Remove schedules plan from database and cancel waiting & running
schedules tasks in this repository when default branch changed.
Remove unnecessary syncbranchToDB with tests (#28624) (#28629)
Replace #28625
Backport #28624 by lunny
#28361 introduced `syncBranchToDB` in `CreateNewBranchFromCommit`. This
PR will revert the change because it's unnecessary. Every push will
already be checked by `syncBranchToDB`.
This PR also created a test to ensure it's right.
Also sync DB branches on push if necessary (#28361) (#28403)
Fix #28056
Backport #28361
This PR will check whether the repo has zero branch when pushing a
branch. If that, it means this repository hasn't been synced.
The reason caused that is after user upgrade from v1.20 -> v1.21, he
just push branches without visit the repository user interface. Because
all repositories routers will check whether a branches sync is necessary
but push has not such check.
For every repository, it has two states, synced or not synced. If there
is zero branch for a repository, then it will be assumed as non-sync
state. Otherwise, it's synced state. So if we think it's synced, we just
need to update branch/insert new branch. Otherwise do a full sync. So
that, for every push, there will be almost no extra load added. It's
high performance than yours.
For the implementation, we in fact will try to update the branch first,
if updated success with affect records > 0, then all are done. Because
that means the branch has been in the database. If no record is
affected, that means the branch does not exist in database. So there are
two possibilities. One is this is a new branch, then we just need to
insert the record. Another is the branches haven't been synced, then we
need to sync all the branches into database.
Part of #27065
This reduces the usage of `db.DefaultContext`. I think I've got enough
files for the first PR. When this is merged, I will continue working on
this.
Considering how many files this PR affect, I hope it won't take to long
to merge, so I don't end up in the merge conflict hell.
---------
Co-authored-by: wxiaoguang <wxiaoguang@gmail.com>
Move web/api context related testing function into a separate package (#26859)
Just like `models/unittest`, the testing helper functions should be in a
separate package: `contexttest`
And complete the TODO:
> // TODO: move this function to other packages, because it depends on
"models" package
This PR
- Fix #26093. Replace `time.Time` with `timeutil.TimeStamp`
- Fix #26135. Add missing `xorm:"extends"` to `CountLFSMetaObject` for
LFS meta object query
- Add a unit test for LFS meta object garbage collection
Add context parameter to some database functions (#26055)
To avoid deadlock problem, almost database related functions should be
have ctx as the first parameter.
This PR do a refactor for some of these functions.
Before: the concept "Content string" is used everywhere. It has some
problems:
1. Sometimes it means "base64 encoded content", sometimes it means "raw
binary content"
2. It doesn't work with large files, eg: uploading a 1G LFS file would
make Gitea process OOM
This PR does the refactoring: use "ContentReader" / "ContentBase64"
instead of "Content"
This PR is not breaking because the key in API JSON is still "content":
`` ContentBase64 string `json:"content"` ``
Do not "guess" the file encoding/BOM when using API to upload files (#25828)
Related issue: #18368
It doesn't seem right to "guess" the file encoding/BOM when using API to
upload files.
The API should save the uploaded content as-is.
Remove unnecessary `if opts.Logger != nil` checks.
* For "CLI doctor" mode, output to the console's "logger.Info".
* For "Web Task" mode, output to the default "logger.Debug", to avoid
flooding the server's log in a busy production instance.
Co-authored-by: Giteabot <teabot@gitea.io>
Fix branch commit message too long problem (#25588)
When branch's commit CommitMessage is too long, the column maybe too
short.(TEXT 16K for mysql).
This PR will fix it to only store the summary because these message will
only show on branch list or possible future search?
Related #14180
Related #25233
Related #22639
Close #19786
Related #12763
This PR will change all the branches retrieve method from reading git
data to read database to reduce git read operations.
- [x] Sync git branches information into database when push git data
- [x] Create a new table `Branch`, merge some columns of `DeletedBranch`
into `Branch` table and drop the table `DeletedBranch`.
- [x] Read `Branch` table when visit `code` -> `branch` page
- [x] Read `Branch` table when list branch names in `code` page dropdown
- [x] Read `Branch` table when list git ref compare page
- [x] Provide a button in admin page to manually sync all branches.
- [x] Sync branches if repository is not empty but database branches are
empty when visiting pages with branches list
- [x] Use `commit_time desc` as the default FindBranch order by to keep
consistent as before and deleted branches will be always at the end.
---------
Co-authored-by: Jason Song <i@wolfogre.com>
1. The "web" package shouldn't depends on "modules/context" package,
instead, let each "web context" register themselves to the "web"
package.
2. The old Init/Free doesn't make sense, so simplify it
* The ctx in "Init(ctx)" is never used, and shouldn't be used that way
* The "Free" is never called and shouldn't be called because the SSPI
instance is shared
---------
Co-authored-by: Giteabot <teabot@gitea.io>
Remove duplicated functions when deleting a branch (#25128)
Extract from #22743
`DeleteBranch` will trigger a push update event, so that
`pull_service.CloseBranchPulls` has been invoked twice and
`AddDeletedBranch` is better to be moved to push update then even user
delete a branch via git command, it will also be triggered.
Co-authored-by: Giteabot <teabot@gitea.io>
API endpoint for changing/creating/deleting multiple files (#24887)
This PR creates an API endpoint for creating/updating/deleting multiple
files in one API call similar to the solution provided by
[GitLab](https://docs.gitlab.com/ee/api/commits.html#create-a-commit-with-multiple-files-and-actions).
To archive this, the CreateOrUpdateRepoFile and DeleteRepoFIle functions
in files service are unified into one function supporting multiple files
and actions.
Resolves #14619
Before there was a "graceful function": RunWithShutdownFns, it's mainly
for some modules which doesn't support context.
The old queue system doesn't work well with context, so the old queues
need it.
After the queue refactoring, the new queue works with context well, so,
use Golang context as much as possible, the `RunWithShutdownFns` could
be removed (replaced by RunWithCancel for context cancel mechanism), the
related code could be simplified.
This PR also fixes some legacy queue-init problems, eg:
* typo : archiver: "unable to create codes indexer queue" => "unable to
create repo-archive queue"
* no nil check for failed queues, which causes unfriendly panic
After this PR, many goroutines could have better display name:
Use the type RefName for all the needed places and fix pull mirror sync bugs (#24634)
This PR replaces all string refName as a type `git.RefName` to make the
code more maintainable.
Fix #15367
Replaces #23070
It also fixed a bug that tags are not sync because `git remote --prune
origin` will not remove local tags if remote removed.
We in fact should use `git fetch --prune --tags origin` but not `git
remote update origin` to do the sync.
Some answer from ChatGPT as ref.
> If the git fetch --prune --tags command is not working as expected,
there could be a few reasons why. Here are a few things to check:
>
>Make sure that you have the latest version of Git installed on your
system. You can check the version by running git --version in your
terminal. If you have an outdated version, try updating Git and see if
that resolves the issue.
>
>Check that your Git repository is properly configured to track the
remote repository's tags. You can check this by running git config
--get-all remote.origin.fetch and verifying that it includes
+refs/tags/*:refs/tags/*. If it does not, you can add it by running git
config --add remote.origin.fetch "+refs/tags/*:refs/tags/*".
>
>Verify that the tags you are trying to prune actually exist on the
remote repository. You can do this by running git ls-remote --tags
origin to list all the tags on the remote repository.
>
>Check if any local tags have been created that match the names of tags
on the remote repository. If so, these local tags may be preventing the
git fetch --prune --tags command from working properly. You can delete
local tags using the git tag -d command.
---------
Co-authored-by: delvh <dev.lh@web.de>
# ⚠️ 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:
There was only one `IsRepositoryExist` function, it did: `has && isDir`
However it's not right, and it would cause 500 error when creating a new
repository if the dir exists.
Then, it was changed to `has || isDir`, it is still incorrect, it
affects the "adopt repo" logic.
To make the logic clear:
* IsRepositoryModelOrDirExist
* IsRepositoryModelExist
Jason Song
Giteabot
Mihir Joshi
Lunny Xiao
6543
JakobDev
Nanguan Lin
wxiaoguang
Zettat123
Chongyi Zheng
yp05327
delvh
silverwind
Denys Konovalov
Yarden Shoham
