Testing multiple zebrads #80
Conversation
|
|
||
| #[ignore = "added to test multiple zcashd instances"] | ||
| #[tokio::test] | ||
| async fn launch_zebrad_multiple_times_with_cache() { |
There was a problem hiding this comment.
This is a sequential launch of zebrad's. Notice how there are being awaited.
We need a single test to launch 2+ instances in parallel. In as comparable as possible of a way as how nextest run launches 2 tests and each test launches a zebrad process.
Only then will they be really competing for the cache dir lock.
There was a problem hiding this comment.
I'm not sure. Are you saying that...
| async fn launch_zebrad_multiple_times_with_cache() { | ||
| tracing_subscriber::fmt().init(); | ||
|
|
||
| let zebrad1 = Zebrad::launch(ZebradConfig { |
There was a problem hiding this comment.
This zebrad1 being launched ...
| chain_cache: Some(utils::chain_cache_dir().join("client_rpc_tests_large")), | ||
| network: network::Network::Regtest, | ||
| }) | ||
| .await |
| zebrad1.print_stdout(); | ||
| zebrad1.print_stderr(); | ||
|
|
||
| let zebrad2 = Zebrad::launch(ZebradConfig { |
There was a problem hiding this comment.
stops because this zebrad2 launching with the same chain_cache config..
| zebrad2.print_stderr(); | ||
|
|
||
| // Don't stop either zebrad | ||
| assert_eq!(zebrad1.get_chain_height().await, 52.into()); |
There was a problem hiding this comment.
and means this test running zebrad1 ..
|
|
||
| // Don't stop either zebrad | ||
| assert_eq!(zebrad1.get_chain_height().await, 52.into()); | ||
| assert_eq!(zebrad2.get_chain_height().await, 52.into()); |
There was a problem hiding this comment.
Or this one, is corrupted because the launches are being awaited?
I get the error ..IO error: While lock file... with launch_zebrad_multiple_times_with_cache and either
launch_zebrad_with_cache or launch_zebrad_with_cache_again
though I have not completely isolated the chain_cache dir yet, I think your fix will make all of these pass.
We can also write a test that will spawn multiple tasks, but I think these tests are useful.
There was a problem hiding this comment.
launch_zebrad_multiple_times_with_cache is failing because zebrad1 isn't being closed. If you do close it, it should launch both alright.
The key aspect is that they only raco for the lock if they are both running at the same time.
launch_zebrad_multiple_times_with_cache runs them sequentially, since its awaiting them.
I agree that it's an useful test. Could be made into 2 distinct ones:
- one that does just this and DOESNT stop
zebrad1and it asserts that it fails (without the copy cache fix) - one that does stop
zebrad1and asserts that both run alright.
BUT this test isn't yet running 2 zebrad instances in parallel like 2 distinct tests each spawning a separate zebrad processes would. That's what we are aiming for.
There was a problem hiding this comment.
I think, in this test, against current dev, it is running two instances at the same time, with the same cache... please read my comment below.
|
Each We could additionally perhaps make new, more robust, longer-running tests similarly. In the case of a single test: all async tasks are managed by the runtime, tokio's runtime.
One other thing we could do is to use I'll make another test where both are launched before any And then, if we really want to get visibility to confirm some state during testing, we could write some kind of script to inspect and record the process(es) running during tests. I added a test with 3 async |
Adding tests that fail with error from #19
Aimed to be fixed with #79