Apport

crashes still appearing regarding os.unlink(uploaded)

Bug #1088587 reported by Brian Murray
6
This bug affects 1 person
Affects Status Importance Assigned to Milestone
Apport
New
Undecided
Unassigned

Bug Description

The following bucket, https://errors.ubuntu.com/bucket/?id=%2Fusr%2Fshare%2Fapport%2Fapport-gtk%3AOSError%3A%3Cmodule%3E%3Arun_argv%3Arun_crashes%3Arun_crash%3Amark_report_upload, is still receiving error reports regarding the version of apport (2.6.1-0ubuntu9 ) in which that specific problem was fixed. For example this problem:

https://errors.ubuntu.com/oops/cea2c0b0-3f72-11e2-8311-e4115b0f8a4a

Has ApportVersion: 2.6.1-0ubuntu9, Package: apport-gtk 2.6.1-0ubuntu9 and the ExecutableTimestamp: 1354705699 matches the new timestamp of the new version of apport-gtk. However, if you look at the Traceback you'll notice that the line numbers correspond to the old version of fileutils.py.

  File "/usr/lib/python3/dist-packages/apport/fileutils.py", line 114, in mark_report_upload
    os.unlink(uploaded)

Looking in Dependencies.txt we can see that python3-apport, which provides fileutils.py, is up to date also. Additionally, the ExecutableTimestamp is less than the Date in the report (when the crash occurred) which if it were not would indicate the package changing after the crash happened.

So, it's really quite mysterious why these are continuing to be reported.

See original description
Revision history for this message
Brian Murray (brian-murray) wrote :

Interestingly some of the reports are missing the second to last line in the Traceback - os.unlink(uploaded)

Revision history for this message
Brian Murray (brian-murray) wrote :

And less helpfully some crash on os.unlink(upload):

https://errors.ubuntu.com/oops/bbbe0350-4180-11e2-8ca3-e4115b0f8a4a

Which I would not expect to be in the same bucket.

Revision history for this message
Brian Murray (brian-murray) wrote :

One of them actually has a different ApportVersion and Package:

https://errors.ubuntu.com/oops/b5afa070-3fcf-11e2-bb40-2c768aafd08c

description: updated
Revision history for this message
Brian Murray (brian-murray) wrote :

Looking at the OOPSes in this bucket - 15 out of 22 (68%) of them have an empty 2nd line and 4 out of 22 (18%) of them were regarding trying to remove .upload.

Steve Langasek (vorlon)
description: updated
Revision history for this message
Steve Langasek (vorlon) wrote :

> Additionally, the ExecutableTimestamp is less than the Date in the report
> (when the crash occurred) which if it were not would indicate the package
> changing after the crash happened.

That's not something you can rely on, because the timestamp of the executable on the system is set to the creation time of the file /within the package/; it's not the time the package was unpacked. So this ordering:

 - new version of package is created
 - running instance of old package crashes
 - new version of package is installed

Can give you a crash timestamp that's later than the on-disk timestamp of the executable even though the package was upgraded after the crash.

That doesn't look like what's happening here, just clarifying.

Revision history for this message
Brian Murray (brian-murray) wrote :

Steve had asked about the version of python3-apport - only one of the reports has python3-apport version 2.6.1-0ubuntu8. That one also has ApportVersion = 2.6.1-0ubuntu8.

http://errors.ubuntu.com/oops/b5afa070-3fcf-11e2-bb40-2c768aafd08c

To post a comment you must log in.
This report contains Public information  
Everyone can see this information.
Subscribing...

Other bug subscribers

Remote bug watches

Bug watches keep track of this bug in other bug trackers.