I believe we have AutoQA 0.4.5 ready:
https://fedorahosted.org/autoqa/milestone/0.4.5
There are a few hot issues remaining, but I don't think we should wait for them:
#293 CmdError: No more mirrors to try - tflink working on this one - can be really difficult - doesn't produce bad results, just sometimes produces no result at all, not blocker
#284 depcheck does not produce output for some pending updates - may be related to the above one - doesn't produce bad results, just sometimes produces no result at all, not blocker
#289 Missing autotest debug logs - this may take some time to debug and solve
So, I want to ask anyone interested to checkout "stable" branch a run a few tests (depcheck and upgradepath are top priorities) and report any problems found. If everything works as expected, we will publish a new release any day now.
Thanks.
I believe we have AutoQA 0.4.5 ready:
https://fedorahosted.org/autoqa/milestone/0.4.5
So, I want to ask anyone interested to checkout "stable" branch a run a few tests (depcheck and upgradepath are top priorities) and report any problems found. If everything works as expected, we will publish a new release any day now.
I haven't found any problems. If neither any of you did, I think we're ready to package and release. James, can you take care of that?
We can enable bodhi comments after deploying to production (today, tomorrow?).
Thanks, Kamil
On Tue, 2011-04-05 at 07:00 -0400, Kamil Paral wrote:
I believe we have AutoQA 0.4.5 ready:
https://fedorahosted.org/autoqa/milestone/0.4.5
So, I want to ask anyone interested to checkout "stable" branch a run a few tests (depcheck and upgradepath are top priorities) and report any problems found. If everything works as expected, we will publish a new release any day now.
I haven't found any problems. If neither any of you did, I think we're ready to package and release.
James, can you take care of that?
Packaged and deployed autoqa-0.4.5-1. Well done everyone!
We can enable bodhi comments after deploying to production (today, tomorrow?).
Is this worth another heads up announcement to maintainers, or is it find just to turn comments back on?
Thanks, James
On 04/05/2011 01:03 PM, James Laska wrote:
On Tue, 2011-04-05 at 07:00 -0400, Kamil Paral wrote:
I believe we have AutoQA 0.4.5 ready:
https://fedorahosted.org/autoqa/milestone/0.4.5
So, I want to ask anyone interested to checkout "stable" branch a run a few tests (depcheck and upgradepath are top priorities) and report any problems found. If everything works as expected, we will publish a new release any day now.
I haven't found any problems. If neither any of you did, I think we're ready to package and release.
James, can you take care of that?
Packaged and deployed autoqa-0.4.5-1. Well done everyone!
We can enable bodhi comments after deploying to production (today, tomorrow?).
Is this worth another heads up announcement to maintainers, or is it find just to turn comments back on?
I think that it would be a good idea to send out a heads-up to devel@ just so that packagers are aware of what's going on.
I'm not expecting more problems, just think that it wouldn't hurt.
Tim
We can enable bodhi comments after deploying to production (today, tomorrow?).
Is this worth another heads up announcement to maintainers, or is it find just to turn comments back on?
Oooh, good idea, I've forgotten. I'll create a short announcement.
As I've been digging into #293 [1] more, I think that we should wait to re-enable bodhi comments until after that has been resolved and pushed to production.
I say this because #293 only shows up when there are accepted builds. autoqa determines a build to be accepted when it still has a -pending tag and has a "passed" autoqa comment in bodhi.
At the moment, only a couple of builds from f13-updates-testing are pending and have bodhi comments because we didn't have comments on for very long. If we turn them on again, we are going to have a lot more builds passed in as "accepted" to depcheck and thus, a lot more tests crashing from #293.
I wanted to release 0.4.5 so that we can enable bodhi comments again, otherwise we could have waited longer. As I described earlier, I don't think #293 is a blocker. It may crash, and it may crash often, true. But if it crashes, it doesn't send any comments, right? So no harm done. If depcheck has 50% (or 80%) crash rate, I don't mind. It's better to have some testing done than no testing done. And upgradepath still should run fine.
It's possible that we discover some further problems in Bodhi/depcheck/upgradepath, so it's better to deploy now and find them out immediately than wait weeks for #293 and discover them only after that.
What do you think?
I wanted to release 0.4.5 so that we can enable bodhi comments again, otherwise we could have waited longer. As I described earlier, I don't think #293 is a blocker. It may crash, and it may crash often, true. But if it crashes, it doesn't send any comments, right? So no harm done. If depcheck has 50% (or 80%) crash rate, I don't mind. It's better to have some testing done than no testing done. And upgradepath still should run fine.
It's possible that we discover some further problems in Bodhi/depcheck/upgradepath, so it's better to deploy now and find them out immediately than wait weeks for #293 and discover them only after that.
What do you think?
Only after writing this reply I've read the emails in autoqa-devel and found out that solution to #293 might be a simple one-line patch. Sorry for that. In this case it makes perfect sense to wait a few more days, test the fix, deploy 0.4.6 and enable bodhi comments only then. My only concern is that jlaska may not be happy about the increased work load :/
On Wed, 2011-04-06 at 03:35 -0400, Kamil Paral wrote:
I wanted to release 0.4.5 so that we can enable bodhi comments again, otherwise we could have waited longer. As I described earlier, I don't think #293 is a blocker. It may crash, and it may crash often, true. But if it crashes, it doesn't send any comments, right? So no harm done. If depcheck has 50% (or 80%) crash rate, I don't mind. It's better to have some testing done than no testing done. And upgradepath still should run fine.
It's possible that we discover some further problems in Bodhi/depcheck/upgradepath, so it's better to deploy now and find them out immediately than wait weeks for #293 and discover them only after that.
What do you think?
Only after writing this reply I've read the emails in autoqa-devel and found out that solution to #293 might be a simple one-line patch. Sorry for that. In this case it makes perfect sense to wait a few more days, test the fix, deploy 0.4.6 and enable bodhi comments only then.
My only concern is that jlaska may not be happy about the increased work load :/
Not at all ... whatever you guys need on the release side. I can handle building + deploying a new version about once a week. Well, I can certainly do more, but I need to reserve time for other activities. :)
I'll stand-by awaiting orders for 0.4.6 and leave bodhi comments disabled.
Thanks all! James
On 04/06/2011 01:35 AM, Kamil Paral wrote:
I wanted to release 0.4.5 so that we can enable bodhi comments again, otherwise we could have waited longer. As I described earlier, I don't think #293 is a blocker. It may crash, and it may crash often, true. But if it crashes, it doesn't send any comments, right? So no harm done. If depcheck has 50% (or 80%) crash rate, I don't mind. It's better to have some testing done than no testing done. And upgradepath still should run fine.
It's possible that we discover some further problems in Bodhi/depcheck/upgradepath, so it's better to deploy now and find them out immediately than wait weeks for #293 and discover them only after that.
What do you think?
Only after writing this reply I've read the emails in autoqa-devel and found out that solution to #293 might be a simple one-line patch. Sorry for that. In this case it makes perfect sense to wait a few more days, test the fix, deploy 0.4.6 and enable bodhi comments only then. My only concern is that jlaska may not be happy about the increased work load :/
Yeah, several hours for a one-line patch. I hope I'm not being evaluated on LOC/hour :)
I'll focus on testing this more today so that we can push to stable and release.
I just thought of this after sending my last email, but it might not be a bad idea to enable comments now if nobody is reading results. As long as we're planning to release a fix for #293 in the next couple of days, I don't think that anyone is going to miss some depcheck runs.
How about this: we enable bodhi comments on Monday. That way, we don't have to worry about any issues over the weekend. If the release for #293 has been deployed, that's good but not a blocker.
Tim
----- Original Message -----
From: "Tim Flink" tflink@redhat.com
<snip /> How about this: we enable bodhi comments on Monday. That way, we don't have to worry about any issues over the weekend. If the release for #293 has been deployed, that's good but not a blocker.
Tim
+1
On 04/06/2011 01:23 AM, Kamil Paral wrote:
We can enable bodhi comments after deploying to production (today, tomorrow?).
Is this worth another heads up announcement to maintainers, or is it find just to turn comments back on?
Oooh, good idea, I've forgotten. I'll create a short announcement.
As I've been digging into #293 [1] more, I think that we should wait to re-enable bodhi comments until after that has been resolved and pushed to production.
I say this because #293 only shows up when there are accepted builds. autoqa determines a build to be accepted when it still has a -pending tag and has a "passed" autoqa comment in bodhi.
At the moment, only a couple of builds from f13-updates-testing are pending and have bodhi comments because we didn't have comments on for very long. If we turn them on again, we are going to have a lot more builds passed in as "accepted" to depcheck and thus, a lot more tests crashing from #293.
I wanted to release 0.4.5 so that we can enable bodhi comments again, otherwise we could have waited longer. As I described earlier, I don't think #293 is a blocker. It may crash, and it may crash often, true. But if it crashes, it doesn't send any comments, right? So no harm done. If depcheck has 50% (or 80%) crash rate, I don't mind. It's better to have some testing done than no testing done. And upgradepath still should run fine.
It's possible that we discover some further problems in Bodhi/depcheck/upgradepath, so it's better to deploy now and find them out immediately than wait weeks for #293 and discover them only after that.
What do you think?
I guess I'm a little more on the cautious side but the question remains "are people actually looking at the results without bodhi comments?".
I agree that some testing is better than none, but I'm not sure that sacrificing almost all of our depcheck runs in the name of getting a feature deployed is the best course. Then again, if nobody outside of our team is actually reading the depcheck results, we're not exactly losing anything since no erroneous comments will be posted.
So I guess that I'm pretty much on the fence for this. I think that you all have a better feel for how the results are being used better than I do.
If nobody is reading the depcheck results now, go ahead and enable bodhi comments.
If people are actually reading them, lets hold off a couple of days and see if we can't get the fix for #293 ready for release. If the fix takes more than a couple of days, just enable comments. Like you said, there may be other issues that we need to solve.
Looking at your other email, it sounds like you are pretty much of the same mind on this, though.
Tim
On 04/04/2011 06:03 AM, Kamil Paral wrote:
I believe we have AutoQA 0.4.5 ready:
https://fedorahosted.org/autoqa/milestone/0.4.5
There are a few hot issues remaining, but I don't think we should wait for them:
#293 CmdError: No more mirrors to try - tflink working on this one - can be really difficult - doesn't produce bad results, just sometimes produces no result at all, not blocker
#284 depcheck does not produce output for some pending updates - may be related to the above one - doesn't produce bad results, just sometimes produces no result at all, not blocker
#289 Missing autotest debug logs - this may take some time to debug and solve
So, I want to ask anyone interested to checkout "stable" branch a run a few tests (depcheck and upgradepath are top priorities) and report any problems found. If everything works as expected, we will publish a new release any day now.
As I've been digging into #293 [1] more, I think that we should wait to re-enable bodhi comments until after that has been resolved and pushed to production.
I say this because #293 only shows up when there are accepted builds. autoqa determines a build to be accepted when it still has a -pending tag and has a "passed" autoqa comment in bodhi.
At the moment, only a couple of builds from f13-updates-testing are pending and have bodhi comments because we didn't have comments on for very long. If we turn them on again, we are going to have a lot more builds passed in as "accepted" to depcheck and thus, a lot more tests crashing from #293.
Tim
autoqa-devel@lists.fedorahosted.org
-
James Laska -
Josef Skladanka -
Kamil Paral -
Tim Flink