New issue
Advanced search Search tips

Issue 809683 link

Starred by 2 users

Issue metadata

Status: Duplicate
Merged: issue 796199
Owner:
Closed: Feb 2018
Cc:
Components:
EstimatedDays: ----
NextAction: ----
OS: Windows
Pri: 2
Type: Bug



Sign in to add a comment

Webpage with large amount of images (2900+) causes some of the image downloads to be canceled.

Reported by koconnor...@gmail.com, Feb 6 2018

Issue description

UserAgent: Mozilla/5.0 (Windows NT 10.0; Win64; x64; rv:58.0) Gecko/20100101 Firefox/58.0

Example URL:

Steps to reproduce the problem:
1. Open the dev tools network tab.
2. Open a webpage with 2900+ images.
3. Observe that some of the image requests are canceled.

What is the expected behavior?
All images should download.

What went wrong?
Some of the image downloads are canceled.

Did this work before? N/A 

Chrome version: 64.0.3282.140 (Official Build) (64-bit) (cohort: 64_140_win)  Channel: stable
OS Version: 10.0
Flash Version: 

I don't experience this problem in Firefox, Edge, or IE.
 
screenshot.png
50.9 KB View Download
Labels: Needs-Feedback
Can you attach a NetLog per these instructions? Thanks!
https://dev.chromium.org/for-testers/providing-network-details

(My guess is this might be  issue #798977 , but let's get a log first.)
I can't provide a NetLog for the page I'm working with due to work restrictions. I'm going to try to reproduce the problem with a dummy page and attach a NetLog.
Project Member

Comment 3 by sheriffbot@chromium.org, Feb 6 2018

Cc: davidben@chromium.org
Labels: -Needs-Feedback
Thank you for providing more feedback. Adding requester "davidben@chromium.org" to the cc list and removing "Needs-Feedback" label.

For more details visit https://www.chromium.org/issue-tracking/autotriage - Your friendly Sheriffbot
Cc: -davidben@chromium.org
Labels: Needs-Feedback
Labels: Needs-Triage-M64
I wasn't able to reproduce the problem in a simpler setting, but I did realize we recently upgraded our proxy server to use HTTP/2. After disabling HTTP/2 all the images load correctly.
Project Member

Comment 7 by sheriffbot@chromium.org, Feb 7 2018

Cc: davidben@chromium.org
Labels: -Needs-Feedback
Thank you for providing more feedback. Adding requester "davidben@chromium.org" to the cc list and removing "Needs-Feedback" label.

For more details visit https://www.chromium.org/issue-tracking/autotriage - Your friendly Sheriffbot
Cc: -davidben@chromium.org mmenke@chromium.org
Components: -Internals>Network Internals>Network>HTTP2
Labels: Needs-Feedback
Could you try reproducing it in an HTTP/2 setting and get a NetLog?

We throttle requests less over HTTP/2, so it's possible this is then hitting  issue #798977 , while the HTTP/1.1 behavior doesn't hit it? +mmenke
Sure, I actually didn't try HTTP/2 when trying to reproduce the bug. I'll try it and see what happens.
Project Member

Comment 10 by sheriffbot@chromium.org, Feb 7 2018

Cc: davidben@chromium.org
Labels: -Needs-Feedback
Thank you for providing more feedback. Adding requester "davidben@chromium.org" to the cc list and removing "Needs-Feedback" label.

For more details visit https://www.chromium.org/issue-tracking/autotriage - Your friendly Sheriffbot
I was able to reproduce the problem and I see a lot of ERR_CONNECTION_RESET errors in the console log which I didn't see where I originally encountered the problem.

The NetLog is too big so I'm zipping it.
chrome-net-export-log.zip
2.1 MB Download

Comment 12 by b...@chromium.org, Feb 8 2018

Owner: b...@chromium.org

Comment 13 by b...@chromium.org, Feb 14 2018

Mergedinto: 796199
Status: Duplicate (was: Unconfirmed)
Thank you for the NetLog.  Seems like server sends a GOAWAY frame:

t=5575 [st=2615]    HTTP2_SESSION_RECV_GOAWAY
                    --> active_streams = 128
                    --> debug_data = "[0 bytes were stripped]"
                    --> error_code = "0 (NO_ERROR)"
                    --> last_accepted_stream_id = 1999
                    --> unclaimed_streams = 0

Some time after this the connection is closed before data for a number of streams with stream ID smaller than or equal to 1999 could arrive, which is an error.

This is not an issue with Chrome.  It is suspected that the server actually writes the missing data frames to the socket, but then closes the socket in such a way that the kernel discards them.  See related  issue 796199 .
Thanks, we are using NGINX so that makes sense.

Sign in to add a comment