Created attachment 627337 [details]
Description of problem:
I saw this once on a primary anaconda console (see screenshot). Anaconda-tb-* file was created, but it is empty (related or another problem?). The problem was NOT reported by integrated exception reported, I saw it accidentally. If I hadn't been looking at tty1, I wouldn't have known about it (I haven't tried to actually start the installation, so I don't know whether it would have succeeded or not).
Version-Release number of selected component (if applicable):
F18 Beta TC4 DVD i386
Steps to Reproduce:
1. boot DVD
2. look at tty1
Created attachment 627338 [details]
Created attachment 627339 [details]
Created attachment 627340 [details]
Created attachment 627341 [details]
Created attachment 627342 [details]
Created attachment 627343 [details]
Created attachment 627344 [details]
Created attachment 627345 [details]
This might and might not be a Beta blocker. Developer feedback is required whether this negatively influences anaconda or not. But it might infringe this criterion:
The installer must be able to report failures to Bugzilla and local disk, with appropriate information included
There was a crash, but the user wasn't notified nor offered to report it. Does crash in python-meh means that the integrated bug reporting tool stops working?
It's a probably race-condition, I saw it again in after a few boot attempts.
This traceback was caused by some exception with no stack, which is something I have never seen before, but obviously can happen.
Discussed at 2012-10-15 QA meeting acting as a blocker review meeting: http://meetbot.fedoraproject.org/fedora-meeting/2012-10-15/fedora-qa.2012-10-15-15.00.log.txt . Agreed that this is a worrying bug but we need more details on what's going on before we can decide blocker status. I wonder if it's the NM bug kamil seems to be hitting quite regularly, and this is just what happens when it happens *early* (the time he hits it seems to vary from run to run).
Vratislav, please check whether this can be related to bug 866434 (recent regression in Beta TC4). I haven't seen this issue in TC3 and prior.
Also, can you tell us whether this is something that could disable the internal bug reporting tool (i.e. a dialog would not pop up for any following errors)? How severe this issue is?
(In reply to comment #13)
> Vratislav, please check whether this can be related to bug 866434 (recent
> regression in Beta TC4). I haven't seen this issue in TC3 and prior.
I think it is not related, because bug 866434 is about traceback that has stack and is correctly handled by python-meh. This bug is however about some traceback, that has no stack at all (could be division by zero or something like that).
> Also, can you tell us whether this is something that could disable the
> internal bug reporting tool (i.e. a dialog would not pop up for any
> following errors)? How severe this issue is?
Python defends itself by disabling custom exception handler that raises exception when handling another one. So if a traceback with no stack appears and python-meh raises exception reported in this bug, Python no longer uses it for exception handling.
Please try this updates image:
(In reply to comment #14)
> > Also, can you tell us whether this is something that could disable the
> > internal bug reporting tool (i.e. a dialog would not pop up for any
> > following errors)? How severe this issue is?
> Python defends itself by disabling custom exception handler that raises
> exception when handling another one. So if a traceback with no stack appears
> and python-meh raises exception reported in this bug, Python no longer uses
> it for exception handling.
In other words, once this issue happens, no exception dialog is popped up, and also it prevents the dialog from appearing for any future exceptions. That means it really hits the aforementioned criterion.
(In reply to comment #15)
> Please try this updates image:
This fixes the issue, a GUI dialog appears with exception with no stack. I also verified that exceptions with stack did not regress, they display as usual.
Discussed at 2012-10-17 blocker review meeting: http://meetbot.fedoraproject.org/fedora-qa/2012-10-17/f18beta-blocker-review-4.2012-10-17-16.00.log.txt . We agreed this is a conditional violation of the criterion "The installer must be able to report failures to Bugzilla and local disk, with appropriate information included" - when an exception with no stack occurs, it breaks the crash reporting functionality. In the end we decided this likely isn't occurring enough to strictly constitute a blocker (we applied the 'would we hold release for this if today was the go/no-go and this was the last bug' test, and felt this is the kind of issue that would probably get fudged). It is rejected as a blocker but accepted as NTH as anything that impacts crash reporting is significant for beta test purposes.
python-meh-0.18-1.fc18 has been submitted as an update for Fedora 18.
* should fix your issue,
* was pushed to the Fedora 18 testing repository,
* should be available at your local mirror within two days.
Update it with:
# su -c 'yum update --enablerepo=updates-testing python-meh-0.18-1.fc18'
as soon as you are able to.
Please go to the following url:
then log in and leave karma (feedback).
python-meh-0.18-1.fc18 pushed stable, closing.