Rewrite queue (#24505)

# ⚠️ 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:

![image](https://user-images.githubusercontent.com/2114189/236620635-55576955-f95d-4810-b12f-879026a3afdf.png)

docs/content/doc/administration/config-cheat-sheet.en-us.md

@@ -89,10 +89,6 @@ In addition there is _`StaticRootPath`_ which can be set as a built-in at build
 - `DEFAULT_PUSH_CREATE_PRIVATE`: **true**: Default private when creating a new repository with push-to-create.
 - `MAX_CREATION_LIMIT`: **-1**: Global maximum creation limit of repositories per user,
    `-1` means no limit.
-- `PULL_REQUEST_QUEUE_LENGTH`: **1000**: Length of pull request patch test queue, make it. **DEPRECATED** use `LENGTH` in `[queue.pr_patch_checker]`.
-   as large as possible. Use caution when editing this value.
-- `MIRROR_QUEUE_LENGTH`: **1000**: Patch test queue length, increase if pull request patch
-   testing starts hanging. **DEPRECATED** use `LENGTH` in `[queue.mirror]`.
 - `PREFERRED_LICENSES`: **Apache License 2.0,MIT License**: Preferred Licenses to place at
    the top of the list. Name must match file name in options/license or custom/options/license.
 - `DISABLE_HTTP_GIT`: **false**: Disable the ability to interact with repositories over the
@@ -465,11 +461,6 @@ relation to port exhaustion.
 - `ISSUE_INDEXER_CONN_STR`: ****: Issue indexer connection string, available when ISSUE_INDEXER_TYPE is elasticsearch, or meilisearch. i.e. http://elastic:changeme@localhost:9200
 - `ISSUE_INDEXER_NAME`: **gitea_issues**: Issue indexer name, available when ISSUE_INDEXER_TYPE is elasticsearch
 - `ISSUE_INDEXER_PATH`: **indexers/issues.bleve**: Index file used for issue search; available when ISSUE_INDEXER_TYPE is bleve and elasticsearch. Relative paths will be made absolute against _`AppWorkPath`_.
-- The next 4 configuration values are deprecated and should be set in `queue.issue_indexer` however are kept for backwards compatibility:
-- `ISSUE_INDEXER_QUEUE_TYPE`: **levelqueue**: Issue indexer queue, currently supports:`channel`, `levelqueue`, `redis`. **DEPRECATED** use settings in `[queue.issue_indexer]`.
-- `ISSUE_INDEXER_QUEUE_DIR`: **queues/common**: When `ISSUE_INDEXER_QUEUE_TYPE` is `levelqueue`, this will be the path where the queue will be saved. **DEPRECATED** use settings in `[queue.issue_indexer]`. Relative paths will be made absolute against `%(APP_DATA_PATH)s`.
-- `ISSUE_INDEXER_QUEUE_CONN_STR`: **addrs=127.0.0.1:6379 db=0**: When `ISSUE_INDEXER_QUEUE_TYPE` is `redis`, this will store the redis connection string. When `ISSUE_INDEXER_QUEUE_TYPE` is `levelqueue`, this is a directory or additional options of the form `leveldb://path/to/db?option=value&....`, and overrides `ISSUE_INDEXER_QUEUE_DIR`. **DEPRECATED** use settings in `[queue.issue_indexer]`.
-- `ISSUE_INDEXER_QUEUE_BATCH_NUMBER`: **20**: Batch queue number. **DEPRECATED** use settings in `[queue.issue_indexer]`.
 
 - `REPO_INDEXER_ENABLED`: **false**: Enables code search (uses a lot of disk space, about 6 times more than the repository size).
 - `REPO_INDEXER_TYPE`: **bleve**: Code search engine type, could be `bleve` or `elasticsearch`.
@@ -480,7 +471,6 @@ relation to port exhaustion.
 - `REPO_INDEXER_INCLUDE`: **empty**: A comma separated list of glob patterns (see https://github.com/gobwas/glob) to **include** in the index. Use `**.txt` to match any files with .txt extension. An empty list means include all files.
 - `REPO_INDEXER_EXCLUDE`: **empty**: A comma separated list of glob patterns (see https://github.com/gobwas/glob) to **exclude** from the index. Files that match this list will not be indexed, even if they match in `REPO_INDEXER_INCLUDE`.
 - `REPO_INDEXER_EXCLUDE_VENDORED`: **true**: Exclude vendored files from index.
-- `UPDATE_BUFFER_LEN`: **20**: Buffer length of index request. **DEPRECATED** use settings in `[queue.issue_indexer]`.
 - `MAX_FILE_SIZE`: **1048576**: Maximum size in bytes of files to be indexed.
 - `STARTUP_TIMEOUT`: **30s**: If the indexer takes longer than this timeout to start - fail. (This timeout will be added to the hammer time above for child processes - as bleve will not start until the previous parent is shutdown.) Set to -1 to never timeout.
 
@@ -488,23 +478,14 @@ relation to port exhaustion.
 
 Configuration at `[queue]` will set defaults for queues with overrides for individual queues at `[queue.*]`. (However see below.)
 
-- `TYPE`: **persistable-channel**: General queue type, currently support: `persistable-channel` (uses a LevelDB internally), `channel`, `level`, `redis`, `dummy`
-- `DATADIR`: **queues/**: Base DataDir for storing persistent and level queues. `DATADIR` for individual queues can be set in `queue.name` sections but will default to `DATADIR/`**`common`**. (Previously each queue would default to `DATADIR/`**`name`**.) Relative paths will be made absolute against `%(APP_DATA_PATH)s`.
-- `LENGTH`: **20**: Maximal queue size before channel queues block
+- `TYPE`: **level**: General queue type, currently support: `level` (uses a LevelDB internally), `channel`, `redis`, `dummy`. Invalid types are treated as `level`.
+- `DATADIR`: **queues/common**: Base DataDir for storing level queues. `DATADIR` for individual queues can be set in `queue.name` sections. Relative paths will be made absolute against `%(APP_DATA_PATH)s`.
+- `LENGTH`: **100**: Maximal queue size before channel queues block
 - `BATCH_LENGTH`: **20**: Batch data before passing to the handler
-- `CONN_STR`: **redis://127.0.0.1:6379/0**: Connection string for the redis queue type. Options can be set using query params. Similarly LevelDB options can also be set using: **leveldb://relative/path?option=value** or **leveldb:///absolute/path?option=value**, and will override `DATADIR`
+- `CONN_STR`: **redis://127.0.0.1:6379/0**: Connection string for the redis queue type. Options can be set using query params. Similarly, LevelDB options can also be set using: **leveldb://relative/path?option=value** or **leveldb:///absolute/path?option=value**, and will override `DATADIR`
 - `QUEUE_NAME`: **_queue**: The suffix for default redis and disk queue name. Individual queues will default to **`name`**`QUEUE_NAME` but can be overridden in the specific `queue.name` section.
-- `SET_NAME`: **_unique**: The suffix that will be added to the default redis and disk queue `set` name for unique queues. Individual queues will default to
- **`name`**`QUEUE_NAME`_`SET_NAME`_ but can be overridden in the specific `queue.name` section.
-- `WRAP_IF_NECESSARY`: **true**: Will wrap queues with a timeoutable queue if the selected queue is not ready to be created - (Only relevant for the level queue.)
-- `MAX_ATTEMPTS`: **10**: Maximum number of attempts to create the wrapped queue
-- `TIMEOUT`: **GRACEFUL_HAMMER_TIME + 30s**: Timeout the creation of the wrapped queue if it takes longer than this to create.
-- Queues by default come with a dynamically scaling worker pool. The following settings configure this:
-- `WORKERS`: **0**: Number of initial workers for the queue.
+- `SET_NAME`: **_unique**: The suffix that will be added to the default redis and disk queue `set` name for unique queues. Individual queues will default to **`name`**`QUEUE_NAME`_`SET_NAME`_ but can be overridden in the specific `queue.name` section.
 - `MAX_WORKERS`: **10**: Maximum number of worker go-routines for the queue.
-- `BLOCK_TIMEOUT`: **1s**: If the queue blocks for this time, boost the number of workers - the `BLOCK_TIMEOUT` will then be doubled before boosting again whilst the boost is ongoing.
-- `BOOST_TIMEOUT`: **5m**: Boost workers will timeout after this long.
-- `BOOST_WORKERS`: **1**: This many workers will be added to the worker pool if there is a boost.
 
 Gitea creates the following non-unique queues:
 
@@ -522,21 +503,6 @@ And the following unique queues:
 - `mirror`
 - `pr_patch_checker`
 
-Certain queues have defaults that override the defaults set in `[queue]` (this occurs mostly to support older configuration):
-
-- `[queue.issue_indexer]`
-  - `TYPE` this will default to `[queue]` `TYPE` if it is set but if not it will appropriately convert `[indexer]` `ISSUE_INDEXER_QUEUE_TYPE` if that is set.
-  - `LENGTH` will default to `[indexer]` `UPDATE_BUFFER_LEN` if that is set.
-  - `BATCH_LENGTH` will default to `[indexer]` `ISSUE_INDEXER_QUEUE_BATCH_NUMBER` if that is set.
-  - `DATADIR` will default to `[indexer]` `ISSUE_INDEXER_QUEUE_DIR` if that is set.
-  - `CONN_STR` will default to `[indexer]` `ISSUE_INDEXER_QUEUE_CONN_STR` if that is set.
-- `[queue.mailer]`
-  - `LENGTH` will default to **100** or whatever `[mailer]` `SEND_BUFFER_LEN` is.
-- `[queue.pr_patch_checker]`
-  - `LENGTH` will default to **1000** or whatever `[repository]` `PULL_REQUEST_QUEUE_LENGTH` is.
-- `[queue.mirror]`
-  - `LENGTH` will default to **1000** or whatever `[repository]` `MIRROR_QUEUE_LENGTH` is.
-
 ## Admin (`admin`)
 
 - `DEFAULT_EMAIL_NOTIFICATIONS`: **enabled**: Default configuration for email notifications for users (user configurable). Options: enabled, onmention, disabled

docs/content/doc/administration/config-cheat-sheet.zh-cn.md

@@ -43,7 +43,6 @@ menu:
 - `DEFAULT_PRIVATE`: 默认创建的git工程为私有。 可以是`last`, `private` 或 `public`。默认值是 `last`表示用户最后创建的Repo的选择。
 - `DEFAULT_PUSH_CREATE_PRIVATE`: **true**:  通过 ``push-to-create`` 方式创建的仓库是否默认为私有仓库.
 - `MAX_CREATION_LIMIT`: 全局最大每个用户创建的git工程数目， `-1` 表示没限制。
-- `PULL_REQUEST_QUEUE_LENGTH`: 小心：合并请求测试队列的长度，尽量放大。
 
 ### Repository - Release (`repository.release`)
 
@@ -111,10 +110,6 @@ menu:
 - `ISSUE_INDEXER_CONN_STR`: ****: 工单索引连接字符串，仅当 ISSUE_INDEXER_TYPE 为 `elasticsearch` 时有效。例如: http://elastic:changeme@localhost:9200
 - `ISSUE_INDEXER_NAME`: **gitea_issues**: 工单索引名称，仅当 ISSUE_INDEXER_TYPE 为 `elasticsearch` 时有效。
 - `ISSUE_INDEXER_PATH`: **indexers/issues.bleve**: 工单索引文件存放路径，当索引类型为 `bleve` 时有效。
-- `ISSUE_INDEXER_QUEUE_TYPE`: **levelqueue**: 工单索引队列类型，当前支持 `channel`， `levelqueue` 或 `redis`。
-- `ISSUE_INDEXER_QUEUE_DIR`: **indexers/issues.queue**: 当 `ISSUE_INDEXER_QUEUE_TYPE` 为 `levelqueue` 时，保存索引队列的磁盘路径。
-- `ISSUE_INDEXER_QUEUE_CONN_STR`: **addrs=127.0.0.1:6379 db=0**: 当 `ISSUE_INDEXER_QUEUE_TYPE` 为 `redis` 时，保存Redis队列的连接字符串。
-- `ISSUE_INDEXER_QUEUE_BATCH_NUMBER`: **20**: 队列处理中批量提交数量。
 
 - `REPO_INDEXER_ENABLED`: **false**: 是否启用代码搜索（启用后会占用比较大的磁盘空间，如果是bleve可能需要占用约6倍存储空间）。
 - `REPO_INDEXER_TYPE`: **bleve**: 代码搜索引擎类型，可以为 `bleve` 或者 `elasticsearch`。
@@ -122,7 +117,6 @@ menu:
 - `REPO_INDEXER_CONN_STR`: ****: 代码搜索引擎连接字符串，当 `REPO_INDEXER_TYPE` 为 `elasticsearch` 时有效。例如： http://elastic:changeme@localhost:9200
 - `REPO_INDEXER_NAME`: **gitea_codes**: 代码搜索引擎的名字，当 `REPO_INDEXER_TYPE` 为 `elasticsearch` 时有效。
 
-- `UPDATE_BUFFER_LEN`: **20**: 代码索引请求的缓冲区长度。
 - `MAX_FILE_SIZE`: **1048576**: 进行解析的源代码文件的最大长度，小于该值时才会索引。
 
 ## Security (`security`)

docs/content/doc/administration/repo-indexer.en-us.md

@@ -30,7 +30,6 @@ Gitea can search through the files of the repositories by enabling this function
 ; ...
 REPO_INDEXER_ENABLED = true
 REPO_INDEXER_PATH = indexers/repos.bleve
-UPDATE_BUFFER_LEN = 20
 MAX_FILE_SIZE = 1048576
 REPO_INDEXER_INCLUDE =
 REPO_INDEXER_EXCLUDE = resources/bin/**