* move shutdownfns, terminatefns and hammerfns out of separate goroutines
Coalesce the shutdownfns etc into a list of functions that get run at shutdown
rather then have them run at goroutines blocked on selects.
This may help reduce the background select/poll load in certain
configurations.
* The LevelDB queues can actually wait on empty instead of polling
Slight refactor to cause leveldb queues to wait on empty instead of polling.
* Shutdown the shadow level queue once it is empty
* Remove bytefifo additional goroutine for readToChan as it can just be run in run
* Remove additional removeWorkers goroutine for workers
* Simplify the AtShutdown and AtTerminate functions and add Channel Flusher
* Add shutdown flusher to CUQ
* move persistable channel shutdown stuff to Shutdown Fn
* Ensure that UPCQ has the correct config
* handle shutdown during the flushing
* reduce risk of race between zeroBoost and addWorkers
* prevent double shutdown
Signed-off-by: Andrew Thornton <art27@cantab.net>
In #15826 it has become apparent that there are a few occasions when a response can
hang during writing, and because there is no timeout go will happily just block
interminably. This PR adds a fixed 5 second timeout to all writes to a connection.
Fix#15826
Signed-off-by: Andrew Thornton <art27@cantab.net>
Fixes http: superfluous response.WriteHeader call from code.gitea.io/gitea/modules/context.(*Response).WriteHeader (response.go:67)
* Looking again we don't need this writeHeader as all of our downstream
implementations will always do it for us
Signed-off-by: Andrew Thornton <art27@cantab.net>
Co-authored-by: Lauris BH <lauris@nix.lv>
Unfortunately some old repositories can have tags with empty Tagger, Commit
or Author. Go-Git variants will always have empty values for these whereas
the native git variant leaves them at nil. The simplest solution is just to
always have these set to empty Signatures.
v156 migration also makes the incorrect assumption that these cannot be empty.
Therefore add some handling to this and add logging and adjust broken
logging elsewhere in this migration.
Signed-off-by: Andrew Thornton <art27@cantab.net>
Co-authored-by: techknowlogick <techknowlogick@gitea.io>
* Decouple TestAction_GetRepoLink and TestSizedAvatarLink.
* Load database for TestCheckGPGUserEmail.
* Load database for TestMakeIDsFromAPIAssigneesToAdd.
* Load database for TestGetUserIDsByNames and TestGetMaileableUsersByIDs.
* Load database for TestUser_ToUser.
* Load database for TestRepository_EditWikiPage.
* Include AppSubURL in test.
* Prevent panic with empty slice.
Co-authored-by: zeripath <art27@cantab.net>
Co-authored-by: techknowlogick <techknowlogick@gitea.io>
* Queue manager FlushAll can loop rapidly - add delay
Add delay within FlushAll to prevent rapid loop when workers are busy
Signed-off-by: Andrew Thornton <art27@cantab.net>
* as per lunny
Signed-off-by: Andrew Thornton <art27@cantab.net>
Co-authored-by: 6543 <6543@obermui.de>
* Move restore repo to internal router and invoke from command to avoid open the same db file or queues files
* Follow @zeripath's review
* set no timeout for resotre repo private request
* make restore repo cancelable
* Use single shared random string generation function
- Replace 3 functions that do the same with 1 shared one
- Use crypto/rand over math/rand for a stronger RNG
- Output only alphanumerical for URL compatibilty
Fixes: #15536
* use const string method
* Update modules/avatar/avatar.go
Co-authored-by: a1012112796 <1012112796@qq.com>
Co-authored-by: a1012112796 <1012112796@qq.com>
It is currenly impossible to detect which "SSO" method is responsible for login. This
PR adds some basic trace logging to these methods.
Signed-off-by: Andrew Thornton <art27@cantab.net>
This PR is another in the vein of queue improvements. It suggests an
exponential backoff for bytefifo queues to reduce the load from queue
polling. This will mostly be useful for redis queues.
Signed-off-by: Andrew Thornton <art27@cantab.net>
Co-authored-by: Lauris BH <lauris@nix.lv>
* Drop back to use IsAnInteractiveSession for SVC
There is an apparent permission change problem when using
IsWindowsService to determine if the SVC manager should be
used.
This PR simply drops back to using IsAnInteractiveSession as
this does not change behaviour.
Fix#15454
Signed-off-by: Andrew Thornton <art27@cantab.net>
* Yes staticcheck I know this is deprecated
Signed-off-by: Andrew Thornton <art27@cantab.net>
* Just leave me alone lint
Signed-off-by: Andrew Thornton <art27@cantab.net>
Co-authored-by: 6543 <6543@obermui.de>
* fix some ui bug about draft release
- should not show draft release in tag list because
it will't create real tag
- still show draft release without tag and commit message
for draft release instead of 404 error
- remove tag load for attachement links because it's useless
Signed-off-by: a1012112796 <1012112796@qq.com>
* add test code
* fix test
That's because has added a new release in relaese test database.
* fix dropdown link for draft release
There was a mistake in #15372 where deferral of gitrepo close occurs before it should.
This PR fixes this.
Signed-off-by: Andrew Thornton <art27@cantab.net>
Since the move to Chi the number of stack frames has proliferated somewhat catastrophically and we're up to 96 frames with multiple tests of the url outside of a trie which is inefficient.
This PR reduces the number of stack frames by 6 through careful use of Route, moves Captcha into its own router so that it only fires on Captcha routes, similarly for avatars and repo-avatars.
The robots.txt, / and apple-touch-icon.png are moved out of requiring Contexter.
It moves access logger higher in the stack frame because there is no reason why it can't be higher.
Extract from #15186
Contains #15292
There is a bug setting the redis db in the common nosql manager whereby the db path
always fails.
This PR fixes this.
Signed-off-by: Andrew Thornton <art27@cantab.net>
There is a potential race in TestChannelQueue_Batch due to boost workers starting up
This PR simply removes the boosts from this test.
Signed-off-by: Andrew Thornton <art27@cantab.net>
* The leveldb shadow queue of a persistable channel queue should always start with 0
workers and just use boost to add additional workers if necessary.
* create a zero boost so that if there are no workers in a pool - boost to start the workers
* actually set timeout appropriately on boosted workers
Signed-off-by: Andrew Thornton <art27@cantab.net>
* Improve performance when there are multiple commits in the last commit cache
* read refs directly if we can
Signed-off-by: Andrew Thornton <art27@cantab.net>
#15546 doesn't completely fix this problem because the error returned is an ObjectNotExist
error not a BranchNotExist error.
Add test for ErrObjectNotExist too
Fix#15257
Signed-off-by: Andrew Thornton <art27@cantab.net>
* If the default branch is not present do not report error on stats indexing
Fix#15257
Signed-off-by: Andrew Thornton <art27@cantab.net>
* as per lunny
Signed-off-by: Andrew Thornton <art27@cantab.net>
Co-authored-by: techknowlogick <techknowlogick@gitea.io>
Fix#15236
* Do not do 40byte conversion within ParseTreeLine
* Missed a to40ByteSHA
Signed-off-by: Andrew Thornton <art27@cantab.net>
Co-authored-by: Andrew Thornton <art27@cantab.net>
zeripath