[MINOR][Web UI] Correctly convert bytes in web UI

[jerryshao]

What changes were proposed in this pull request?

should be 1024 based, not 1000.

How was this patch tested?

Manually verified.

[srowen]

KB = 1000 bytes, KiB = 1024 bytes. According to the suffixes we're using, 1000 is correct at the moment. Is the display inconsistent with something else in the UI or logs?

[jerryshao]

Because in the log it shows Memory MB in 1024 based, while in the web UI it is 1000 based, so this is slightly different.

You could check Utils#bytesToString. I think we should unify this.

[srowen]

Ugh, yeah that's wrong in the sense that we are not showing MB, but MiB. I'd favor fixing the labels here and in Utils.bytesToString?

Then again, I see that we will also parse input of "500kb" as if it's "500KiB", using 1024 not 1000. That's wrong too really. But fixing it means a bit of a behavior change. We can support "500KB" or "500kb" but it would now mean 500_1000 bytes not 500_1024.

Well, maybe best to be consistently wrong than inconsistently wrong. Anyone feel at all inclined to take the hit in behavior change or just leave it?

[jerryshao]

I think here has a precise definition. AFAIK in Spark the conversion is 1024 based either KB, K, or kb, KiB is not so commonly used. And we usually treat everything as 1024 based, so it might not be so necessary to differentiate them.

[srowen]

Meh, it's still ambiguous and there's a defined way to disambiguate, so it's unfortunate, but I'm OK with a step towards consistency in any event.

[SparkQA]

Test build #64410 has finished for PR 14804 at commit fe78ecc.

• This patch fails PySpark unit tests.
• This patch merges cleanly.
• This patch adds no public classes.

[SparkQA]

Test build #72944 has finished for PR 14804 at commit fe78ecc.

• This patch fails Spark unit tests.
• This patch merges cleanly.
• This patch adds no public classes.