ci: set behavior test ci for aliyun drive

[suyanhanx]

No description provided.

[Xuanwo]

aliyun drive test failed for

failures:

---- behavior::test_list_empty_dir ----
Unexpected (persistent) at List::next, context: { service: aliyun_drive, path: / } => The resource file cannot be found. file not exist


failures:
    behavior::test_list_empty_dir

test result: FAILED. 113 passed; 1 failed; 0 ignored; 0 measured; 0 filtered out; finished in 291.46s

[suyanhanx]

aliyun drive test failed for

Thanks. I'm handling this on my local machine.

[suyanhanx]

The env var RUST_TEST_THREADS doesn't seem to be working properly.

[Xuanwo]

I feel like the implemention has some bug to address. For example:

---- behavior::test_list_empty_dir ----
Unexpected (persistent) at List::next, context: { service: aliyun_drive, path: / } => The resource file cannot be found. file not exist

List empty dir should return None instead of not found?

[suyanhanx]

It's superficial. Most of those errors reached the limit of read/write.

[Xuanwo]

As I mentioned, the tests go well when limited to 1 thread:

Odd, we've already set test_threads=1.

opendal/.github/services/aliyun_drive/aliyun_drive/action.yml

Lines 33 to 38 in f95cb4d

	- name: Setup test threads
	shell: bash
	run: |
	cat << EOF >> $GITHUB_ENV
	RUST_TEST_THREADS=1
	EOF

[Xuanwo]

Based on my testing on my host, RUST_TEST_THREADS=1 does not behave the same as --test-threads=1.

But docs said

This can also be specified with the RUST_TEST_THREADS environment variable.

https://doc.rust-lang.org/rustc/tests/index.html#--test-threads-num_threads

[suyanhanx]

It's a bit strange. The env var is also behaving differently on my local machine.

I found this:
rust-lang/rust#104053

[suyanhanx]

[yuchanns]

I rewrite the lister (without getting file details), and LGTM on green. Previously, the test-check would list items while other threads delete things, leading to the random 404 files not found.

[Xuanwo]

Wow, so cool! Is this PR ready to go?

[Xuanwo]

Thanks a lot!