Re: [PATCH 0/1] bcache: reduce front IO latency during GC
Coly Li <[email protected]>
| Newsgroups | org.kernel.vger.linux-bcache |
|---|---|
| Message-ID | <[email protected]> |
Hi Robert, Thanks for the great effort! Let me take a look and respond later. Coly Li > 2025年7月18日 06:52，Robert Pang <[email protected]> 写道： > > Hi Coly > >>> With the min heap regression addressed by the recent revert [1], I am >>> going to initiate a rerun of our last test set [2]. I will share the >>> results as soon as they are available. > > My apologies for the delay in sharing the results due to my work schedule. > > I've completed the rerun of the tests, and the results are available > for your review in [1]. As expected, the data confirms the desired > latency improvements. Furthermore, I've checked for any unintended > side effects and can confirm that no regressions were observed. > > Please take a look and let me know what you think. If there are any > other specific scenarios or tests you would like me to execute to > further validate the changes, please don't hesitate to let me know. > > Best regards > Robert Pang > > [1] https://gist.github.com/robert-pang/a22b7c5dee2600be3260f4db57e5776d#file-test-results-md > > On Thu, Jul 3, 2025 at 9:47 AM Coly Li <[email protected]> wrote: >> >> >> >>> 2025年7月3日 23:34，Coly Li <[email protected]> 写道： >>> >>> Hi Robert, >>> >>>> 2025年7月3日 23:28，Robert Pang <[email protected]> 写道： >>>> >>>> Hi Coly >>>> >>>> I'm writing to provide an update on this bcache GC latency patch. >>>> >>>> With the min heap regression addressed by the recent revert [1], I am >>>> going to initiate a rerun of our last test set [2]. I will share the >>>> results as soon as they are available. >>>> >>>> To ensure this patch is fully qualified for the next merge window, I'm >>>> eager to perform any further tests you deem necessary. What additional >>>> test scenarios would you recommend?" >>>> >>> >>> Can you also do some full random write (4K and 64K block size) performance testing? >>> >>> If there is no performance regression (and I believe improvement will be observed), then let’s go ahead. >> >> BTW, I don’t commit that this patch will go into next merge window. >> >> Thanks. >> >> Coly Li >> >> >> >