Re: [PATCH 00/21] verify fixes and a new test suite

Vincent Fu <[email protected]>
Newsgroups org.kernel.vger.fio
Message-ID <[email protected]>
On 3/5/25 7:36 AM, Shinichiro Kawasaki wrote:
> On Feb 27, 2025 / 11:54, Vincent Fu wrote:
>> On 2/27/25 5:47 AM, Ankit Kumar wrote:
>>> This series fixes few issues with verify, and introduces a new test
>>> suite.
>>>
>>> Inital few patches add the missing client/server support for
>>> verify_write_sequence. This also changes its behavior such that if
>>> verify_write_sequence is explicitly enabled fio will not disable it
>>> under any circumstance.
>>>
>>> Numerous header seed mismatch issues have been reported. This series
>>> introduces a new option verify_header_seed which is similar to
>>> verify_write_sequence, which allow users to disable any header seed
>>> verification. For certain workloads which used to overwrite header seed
>>> before verification, we simply disable the header seed checks now. This
>>> now includes a few more scenarios such as verify_only mode, read only
>>> workloads and workloads with norandommap, where the header seed match is
>>> not guaranteed.
>>>
>>> Few more fixes related to verify_offset, workloads that have offset
>>> modifiers, verification issue when multiple files are specified, are part
>>> of this series.
>>>
>>> Lastly this includes robust test suite for verify.
>>> CI run result: https://github.com/vincentkfu/fio/actions/runs/13552248490
>>>
>>> Note: The fixes in this series doesn't cover experimental_verify and any
>>> workload with variable block sizes.
>>>
>>
>> I have worked closely with Ankit on these patches and hope others can take a
>> look to point out anything we may have missed.
> 
> Hi Ankit, Vincent, thanks for the work and the heads up.
> 
> I ran t/zbd/run-tests-against-nullb with this series, and observed no failure.
> So it looks good from zonemode=zbd testing point of view.
> 
> I took a glance on the changes and did not find anything wrong. It's ok for me
> to apply the changes.

I have applied the patches with small changes to the test script to skip 
a flaky test on macOS and fix a test that did not run reliably on Windows.

Shinichiro, thank you for testing and taking a look at the patches.

Vincent
lmpx.com only provides a reader for public news (NNTP) servers. It is not affiliated with the servers or forums shown here and is not responsible for the content of articles, which is written by their respective authors.