Project file uploads timeout (and often do not OOPS)

[Brad Crittenden]

A user reported problems uploading a 12MB file to the 'freevial' project and receiving no OOPS but instead:

"Please try again
Sorry, there was a problem connecting to the Launchpad server.
Try reloading this page in a minute or two. If the problem persists, let us know in the #launchpad IRC channel on Freenode.
Thanks for your patience."

Tom H. looked at the Apache logs and found:
https://pastebin.canonical.com/2732/

A search for others reporting the problem found:
http://issues.apache.org/bugzilla/show_bug.cgi?id=37770

The Apache server receives the fully uploaded file before passing it on to Launchpad. The upload died before finishing therefore there was no chance for an OOPS to be generated.

OOPS-6acb6414862499859f26aef311dc7bd3

[Joey Stanford (joey)]

Searching the logs, Tom has found a high number of instances of the error in the above pastebin.

There appears to be two issues at work here:

1) A possible apache issue causing (at least) upload failures

2) A lack of apache error log integration with our oops tools which allow us to detect these errors and correct them. Users instead see the unplanned-offline.html file with no OOPS generated to let us know there was actually an issue.

[Christian Reis (kiko)]

See bug #193062 for another case of OOPSes that look like apache timeouts.

[William Grant (wgrant)]

Any updates on this? People are complaining about it fairly often...

[Albert Cervera i Areny - http://www.NaN-tic.com (albert-nan)]

I'm having the same issue uploading a file of 40Mb in openobject-client-kde project. I've uploaded files that large in the past without problems. Anything I can do?

[Curtis Hovey (sinzui)]

A faster connection is about all that can be done, and that is not always an option. There can be connectivity issues between you and launchpad that may resolve themselves.

You could place the file in another public location (http or ftp), then set the filereleaseglob for the series find just the files you want to upload. The product-release-finder runs once a day and looks for files that match the glob. It will match the file name to the milestone and add the file for you.

[Robert Collins (lifeless)]

It may be easiest to fix this by moving uploads out of the zope appservers, which our infrastructure is not geared around supporting big uploads too. E.g. upload directly to the librarian, then register the content with LP (/handwave)

[Abel Deuring (adeuring)]

I could reproduce the problem when I truied to upload an ordinary Linux source code tarball to qastaging, using a somehwaht crappy mobile phone internet connection. The error occured around 1630..1640 UTC, 2011-09-26 (sorry, I did not watch the browser window closely enough to have a more precise time...) Problem: It seems that we don't get an OOPS report.

[Abel Deuring (adeuring)]

next attempt failed 2011-09-26, 1703UTC

[Joey Ye (jinyun-ye)]

Reproduce this bug by uploading a source tarball 130m in size. Time was about 1000UTC 2011-11-30

[Andrew Stubbs (ams-codesourcery)]

I'm getting the same problem trying to upload a 70MB file to the gcc-linaro project.

[Robert Collins (lifeless)]

Andrew succeeded using firefox from a different machine, was failing from chrome. May be irrelevant. Also reports a delay between the browser completing the upload and the error.

[Jānis Kangarooo (kangarooo)]

Maybe finally can timeout made longer waiting?

[Markus Klyver (markusklyver)]

I get the HTTP 502 error message when trying to send >30 MiB bug reports.

[Curtis Hovey (sinzui)]

This post has an example of use modporter to handle uploads: http://blog.jazkarta.com/2010/09/21/handling-large-files-in-plone-with-ore-bigfile/

[iram (mfcv)]

HTTP Error 502: Bad Gateway
While processing ubuntu-bug libinput10 in Jammy