"EOF occurred in violation of protocol" when downloading Python dependencies

RESOLVED WORKSFORME

Status

()

Core
Build Config
--
blocker
RESOLVED WORKSFORME
a year ago
a year ago

People

(Reporter: WG9s, Unassigned)

Tracking

({regression})

Trunk
ARM
Android
regression
Points:
---

Firefox Tracking Flags

(firefox51 unaffected, firefox52 unaffected, firefox53 unaffected, firefox54 affected, firefox55 affected)

Details

(Reporter)

Description

a year ago
I am getting the following errors:

10:09.12 Installed /home/wag/mozilla/mozilla2/testing/mozbase/mozlog
10:09.12 Processing dependencies for mozlog==3.4
10:09.12 Searching for blessings>=1.3
10:09.12 Reading https://pypi.python.org/simple/blessings/
10:09.12 Downloading https://pypi.python.org/packages/af/4a/61acd1c6c29662d3fcbcaee5ba95c20b1d315c5a33534732b6d81e0dc8e8/blessings-1.6.tar.gz#md5=4f552a8ebcd4982693c92571beb99394
10:09.12 error: Download error for https://pypi.python.org/packages/af/4a/61acd1c6c29662d3fcbcaee5ba95c20b1d315c5a33534732b6d81e0dc8e8/blessings-1.6.tar.gz#md5=4f552a8ebcd4982693c92571beb99394: EOF occurred in violation of protocol (_ssl.c:661)
(Reporter)

Updated

a year ago
Summary: Unalbe to build fennec today → Unable to build fennec today
(Reporter)

Comment 1

a year ago
IF this has to do with some new build system requirements this needs to be backed out and re applied with a proper test during the configuration step after the uplift.
https://github.com/kennethreitz/requests/issues/3605 ?
Component: Build Config & IDE Support → Build Config
Product: Firefox for Android → Core
Summary: Unable to build fennec today → "EOF occurred in violation of protocol" when downloading Python dependencies
Version: 54 Branch → Trunk
See also Bug 1318062 and Bug 1039076, which suggest that this might be plain ol' network troubles.
Severity: blocker → normal
(Reporter)

Comment 4

a year ago
(In reply to Richard Newman [:rnewman] from comment #2)
> https://github.com/kennethreitz/requests/issues/3605 ?

Thanks for figuring out how to report this upstream.  I hate things like this that happen on merge day.
Severity: normal → blocker
(Reporter)

Comment 5

a year ago
But then I an not complete a build, so not sure how this is not a blocker.
If it only affects you, it's not a blocker. If it affects everyone, or it affects automation, it's critical/blocker. Remember that the most likely cause of build issues is your setup or your infrastructure -- if it works for the thousands of builds done per day in Mozilla automation and by Mozilla devs, the odds are that something special is happening to you.

With only rare exceptions, component owners set priorities, not reporters.
Severity: blocker → normal
(Reporter)

Comment 7

a year ago
It worked yesterday, it does not work today.
(Reporter)

Comment 8

a year ago
It also works on other platforms like Linux and Windows.
(Reporter)

Comment 9

a year ago
I am re-cloning based on yesterdays changeset and retrying the android build which previously worked just fine.
(Reporter)

Updated

a year ago
Keywords: regression
(Reporter)

Comment 10

a year ago
This is a regression between changeset eb2364853477 and changeset 966464a68a2c.  This has nothing to do with a cahnge in my build configuration.
Severity: normal → blocker
(Reporter)

Comment 12

a year ago
A difference between working and not with the exact same python and every other package installed.
(Reporter)

Comment 13

a year ago
I just don't want whatever cause this just before merge stupid thing that was allowed to land to make it to version 54.
(Reporter)

Comment 14

a year ago
There will be a regressor identified soon.
(In reply to Bill Gianopoulos [:WG9s] from comment #0)
> I am getting the following errors:
> 
> 10:09.12 Installed /home/wag/mozilla/mozilla2/testing/mozbase/mozlog
> 10:09.12 Processing dependencies for mozlog==3.4
> 10:09.12 Searching for blessings>=1.3
> 10:09.12 Reading https://pypi.python.org/simple/blessings/
> 10:09.12 Downloading
> https://pypi.python.org/packages/af/4a/
> 61acd1c6c29662d3fcbcaee5ba95c20b1d315c5a33534732b6d81e0dc8e8/blessings-1.6.
> tar.gz#md5=4f552a8ebcd4982693c92571beb99394
> 10:09.12 error: Download error for
> https://pypi.python.org/packages/af/4a/
> 61acd1c6c29662d3fcbcaee5ba95c20b1d315c5a33534732b6d81e0dc8e8/blessings-1.6.
> tar.gz#md5=4f552a8ebcd4982693c92571beb99394: EOF occurred in violation of
> protocol (_ssl.c:661)

When does this happen? (what command did you run?)
status-firefox51: --- → unaffected
status-firefox52: --- → unaffected
status-firefox53: --- → unaffected
status-firefox55: --- → affected
(Reporter)

Comment 16

a year ago
OK here is the background on this.

I do my own Nightly like builds of Firefox.  On the same system I do a clone based on the nightly changeset, and from the same source tree in sequence I do a 64-bit Linux build an Android build and then a 32-bi Linux build.  The 2 Linux builds worked so sandwiching the filed Android build.  Thefrore I am fairly certain nothing was wrong with the source.

I next tried a depend Android build to see if that would resolve the issue and it did not.  I then tried rebooting the server and doing a clobber build which also failed.  I then filed this bug.

Just to make sure I could actually repeat the success form the day before, I subsequently re-cloned based on the previous days changeset and verified that still build android correctly.  However, it was a couple of hours later before I decided to try that.

Fast forward to today.  Today's builds worked just fine.  Therefore, my best guess is that the file being downloaded from https://pypi.python.org/simple/blessings/ was in fact corrupt.

In any event I am closing this bug.
Status: NEW → RESOLVED
Last Resolved: a year ago
Resolution: --- → WORKSFORME
(Reporter)

Comment 17

a year ago
I was not sure what the correct thing to do with the tracking flags is in a case like this where the bug is resolved as either INVALID or WORKSFORME by the original reporter.
You need to log in before you can comment on or make changes to this bug.