478578 - On Firefox nightly updates, the download speed is significantly miscalculated

Status: RESOLVED WORKSFORME

(Core :: Networking, defect)

Description

[Damian Shaw [Quan]]

STR:

1) On the 1.9.1 Branch, download a software update
2) The download speed is calculated as some very low bytes per second when the download is completed in only a few seconds.

Expected results:

Correct download speed.

Comment 1

[Damian Shaw [Quan]]

Attachment: Screenshot of updater

As you can see here, very low download speed shown and very high estimated time. The download was actually completed in about 5 seconds.

Comment 2

[Robert Strong (they/them - no direct email)]

These values are provided by nsIncrementalDownload and not app update so over to core networking

Comment 3

[Ria Klaassen (not reading all bugmail)]

Don't know if it's related, but yesterday I burnt the contents of my Vista desktop, with lots of firefoxes on it, to a dvd, first it said it would take 20 minutes, but while burning it told me it would take more than 20 days.

Comment 4

[Michael Kohler [:mkohler]]

(In reply to comment #3)
> Don't know if it's related, but yesterday I burnt the contents of my Vista
> desktop, with lots of firefoxes on it, to a dvd, first it said it would take 20
> minutes, but while burning it told me it would take more than 20 days.

I think that is because of the OS and has nothing to do with Firefox.

Comment 5

[WADA:World Anti-bad-Duping Agency]

> Platform: Windows Server 2003
> 1) On the 1.9.1 Branch
> 2) The download speed is calculated as some very low bytes per second when the
download is completed in only a few seconds.

Mismatch between change by Bug 454990 and download manager?
Bug 454990 has increased TCP send window size on Windows, with introducing new prfs.
> +// The default TCP send window on Windows is too small, and autotuning only occurs on receive
> +pref("network.tcp.sendbuffer", 131072);
It's different from (larger than) usual MS Win's default TCP window size. It may be used in receive too, because Bug 454990 changes SocketTransportService.

Comment 6

[Damian Shaw [Quan]]

I didn't get this problem on a full download of Firefox, but then next update was a partial one and I got the problem again then.

Comment 7

[Jukka Pirinen]

I had this same problem when automatic upgrading from Firefox 3.5.4 to 3.5.8 with Windows XP SP3.