aboutsummaryrefslogtreecommitdiffstatshomepage
path: root/src/server.c (unfollow)
Commit message (Collapse)Author
2023-12-12switch to egor@tensin.nameEgor Tensin
2023-11-15implement a command to list runsEgor Tensin
2023-11-12refactoringEgor Tensin
2023-07-18switch to JSON-RPC as message formatEgor Tensin
Instead of the weird `struct msg` I had, I switched to the JSON-RPC format. It's basically the same, but has a well-defined semantics in case of errors.
2023-07-09store process output in SQLiteEgor Tensin
2023-07-08test: verify that added runs are in the databaseEgor Tensin
And that they're marked as finished. It immediately exposed some concurrency bugs, so some locking has been fixed.
2023-07-07server: fix a possible leakEgor Tensin
2023-07-05tcp_server: keep track of client threadsEgor Tensin
This is a major change, obviously; brought to me by Valgrind, which noticed that we don't actually clean up after cimple-client threads. For a more thorough explanation, please see the added comment in tcp_server.c.
2023-07-04sanitize #include-sEgor Tensin
2023-07-04move custom message parsing to a separate moduleEgor Tensin
2023-07-04storage_sqlite: refactoringEgor Tensin
2023-07-04storage: mark completed runs as suchEgor Tensin
2023-07-04storage: requeue old runs from storage on startupEgor Tensin
2023-07-04tcp_server: always clean up connection descriptorsEgor Tensin
2023-07-04sqlite: store new runs in SQLiteEgor Tensin
2023-07-04storage_sqlite: refactoringEgor Tensin
2023-06-13minor refactoringEgor Tensin
2023-06-13signal: remove the stupid add_to_event_loop wrapperEgor Tensin
2023-06-13server: handle disconnected workers gracefullyEgor Tensin
2023-06-13signal: refactoringEgor Tensin
2023-06-13use signalfd to stop on SIGTERMEgor Tensin
Is this an overkill? I don't know. The thing is, correctly intercepting SIGTERM (also SIGINT, etc.) is incredibly tricky. For example, before this commit, my I/O loops in server.c and worker.c were inherently racy. This was immediately obvious if you tried to run the tests. The tests (especially the Valgrind flavour) would run a worker, wait until it prints a "Waiting for a new command" line, and try to kill it using SIGTERM. The problem is, the global_stop_flag check could have already been executed by the worker, and it would hang forever in recv(). The solution seems to be to use signalfd and select()/poll(). I've never used either before, but it seems to work well enough - at least the very same tests pass and don't hang now.
2023-06-11msg: rework some APIsEgor Tensin
2023-05-15signal: refactoringEgor Tensin
2023-05-15minor refactoringEgor Tensin
2023-05-15signal: refactoring, add comments in tcp_server, etc.Egor Tensin
2023-05-15EINVAL means EINTR also?Egor Tensin
2023-05-15rework server-worker communicationEgor Tensin
OK, this is a major rework. * tcp_server: connection threads are not detached anymore, the caller has to clean them up. This was done so that the server can clean up the threads cleanly. * run_queue: simple refactoring, run_queue_entry is called just run now. * server: worker threads are now killed when a run is assigned to a worker. * worker: the connection to server is no longer persistent. A worker sends "new-worker", waits for a task, closes the connection, and when it's done, sends the "complete" message and waits for a new task. This is supposed to improve resilience, since the worker-server connections don't have to be maintained while the worker is doing a CI run.
2023-05-14command: adjust order of parameters to handlersEgor Tensin
2023-05-14msg: add functions for one-off communicationEgor Tensin
2023-05-13ci_queue -> run_queueEgor Tensin
Also, some minor refactoring.
2023-05-13command: refactoringEgor Tensin
2023-05-13best practices & coding style fixesEgor Tensin
* I don't really need to declare all variables at the top of the function anymore. * Default-initialize variables more. * Don't set the output parameter until the object is completely constructed.
2023-05-13add command module to handle request-response communicationsEgor Tensin
2023-05-12ci_queue: rename a couple of functionsEgor Tensin
2023-05-06shut down server/workers gracefully on SIGTERMEgor Tensin
2023-05-06get rid of __attribute__((constructor))Egor Tensin
Explicit is better than implicit.
2023-04-29make struct ci_queue_entry opaqueEgor Tensin
2023-04-29make struct server opaqueEgor Tensin
2023-04-29make struct tcp_server opaqueEgor Tensin
2023-04-27fix a typoEgor Tensin
2023-04-27rename commandsEgor Tensin
2022-12-02add copyright noticesEgor Tensin
2022-09-11create SQLite database on startupEgor Tensin
2022-09-08log: refactoringEgor Tensin
2022-09-08sanitize #include-sEgor Tensin
2022-08-28update command namesEgor Tensin
2022-08-28server: notify workers about requeued jobsEgor Tensin
This allows free workers to pick up jobs after dead workers.
2022-08-28server: notify all threads about shutting downEgor Tensin
The problem is pthread_cond_destroy is unsafe to call if there're threads waiting in pthread_cond_wait. I'm not sure this fix is enough: what if the "broadcast" doesn't reach the threads until we call pthread_cond_destroy? Does it even work that way? Idk
2022-08-28make proper "error" messagesEgor Tensin
Previously, the client had no way to distinguish errors from succesful calls.
2022-08-28make compilers happierEgor Tensin
generated by cgit v1.2.3 (git 2.25.1) at 2024-11-10 00:07:29 +0000