Some files on ftp.mozilla.org after a certain date are erroneously served with `content-type: application/octet-stream` header
Categories
(Release Engineering :: General, defect, P1)
Tracking
(Not tracked)
People
(Reporter: aoia7rz7l, Assigned: hneiva)
Details
(Keywords: regression)
Attachments
(1 file)
I was trying to view the SHASUMS files for 129.0b9 from within the browser but a download prompt popped up instead.
Looking at devtools' netmonitor, it would seem like the server returned content-type: application/octet-stream, when it should have returned content-type: text/plain.
This is reproducible for any file since 129.0b5 but not 129.0b4. For example, the content-type for 129.0b4's Windows installer is application/x-ms-dos-program, but became application/octet-stream in 129.0b5 or later.
Not reproducible when opening or downloading the same files from archive.mozilla.org (which AFAIK is just an alias for ftp.mozilla.org?).
Actually this is now also reproducible on archive.mozilla.org (and releases.mozilla.org). Perhaps some new configuration has propagated?
FWIW nightlies and candidate builds are not affected for some reason. So this is currently affecting at least pub/firefox/releases/, /pub/thunderbird/releases/, and /pub/devedition/releases/ on the file server. However, files under /pub/fenix/releases/ and pub/focus/releases/ seem to be unaffected.
Moving this back to untriaged because I have no idea what the correct component for ftp.mozilla.org bugs should be.
Comment 4•1 year ago
|
||
Trying Release Engineering::General, based on other bugs (e.g. bug 1895965).
Comment 5•1 year ago
|
||
It looks like the content-type (and other headers like cache-control) is lost when the file gets moved from candidates to releases:
$ curl -Is https://archive.mozilla.org/pub/firefox/candidates/129.0b8-candidates/build1/SHA256SUMS | grep content-type
content-type: text/plain
$ curl -Is https://archive.mozilla.org/pub/firefox/releases/129.0b8/SHA256SUMS | grep content-type
content-type: application/octet-stream
https://github.com/mozilla-releng/scriptworker-scripts/pull/1037 landed on July 16, between 129.0b4 and 129.0b5, and affected that part of the process.
Comment 6•1 year ago
|
||
| Assignee | ||
Comment 7•1 year ago
|
||
Fix merged to master and scheduled to land on November 5th. Let me know if this needs to go in before that.
Comment 8•1 year ago
|
||
The severity field is not set for this bug.
:bhearsum, could you have a look please?
For more information, please visit BugBot documentation.
Updated•1 year ago
|
Updated•1 year ago
|
| Assignee | ||
Comment 9•1 year ago
|
||
FYI: The fix is already in production, just waiting for next release cycle (I believe scheduled for Nov 26th).
| Reporter | ||
Comment 10•1 year ago
|
||
I can confirm this issue is no longer reproducible for beta builds since 133.0b5 and release builds since 132.0.2.
Comment 11•1 year ago
|
||
Thanks!
Updated•1 year ago
|
Description
•