Re: Rejected commits to master
Since it was a simple patch, I used Github UI to push the commit. And now I see that Jenkins is broken by it somehow. On Wed, May 22, 2019 at 2:36 PM Julian Hyde wrote: > That sounds plausible. > > It looks as if the system accepted a commit from Laurent on one repo, > and a commit from me in another, while the repos where not in sync. It > sent emails confirming them, then later lost my commit. The CAP > theorem holds - we had availability in the presence of a partition, > but lost consistency. I'd rather sacrifice availability. Or only let > people commit to one of the repos. > > I am worried that Apache INFRA don't realize that a distributed > transactional system is a difficult beast. > > Julian > > On Wed, May 22, 2019 at 2:17 PM Kevin Risden wrote: > > > > I think there have been some delays syncing between github and gitbox. I > > think it depends which origin you are pointing at and which commit is > there. > > > > Since both github and gitbox are masters, we can actually push to either > > one and I would guess there is some checks to make sure commits don't > have > > issues but a pull might be slightly out of date? > > > > Kevin Risden > > > > > > On Wed, May 22, 2019 at 5:12 PM Julian Hyde wrote: > > > > > I pushed a commit to the master branch this morning and it was > > > rejected due to an intervening commit from Laurent. But that > > > intervening commit did not appear when I did 'git fetch origin'. > > > > > > Just now, I was just able to push successfully. In both cases I used > > > 'git push origin master' from the command line, without the '-f' > > > (force) flag. > > > > > > You can see that both commits generated emails: [1] [3]. Laurent's > > > intervening commit also generated an email: [2]. > > > > > > All, Has anyone else seen this behavior? > > > > > > Laurent, What command did you use to commit? > > > > > > Julian > > > > > > [1] > > > > https://lists.apache.org/thread.html/ac70079b65305a9040bfe2ed0720768343d2f6dc038b9a63c081ff0b@%3Ccommits.calcite.apache.org%3E > > > > > > [2] > > > > https://lists.apache.org/thread.html/bddfd2ad8cbe3148820b334b1554fae23d4e05a2b00967b71e5d7191@%3Ccommits.calcite.apache.org%3E > > > > > > [3] > > > > https://lists.apache.org/thread.html/335c042be3fbd042f327e9742d45de790d731d8a7f883294dacf9322@%3Ccommits.calcite.apache.org%3E > > > >
Re: Rejected commits to master
That sounds plausible. It looks as if the system accepted a commit from Laurent on one repo, and a commit from me in another, while the repos where not in sync. It sent emails confirming them, then later lost my commit. The CAP theorem holds - we had availability in the presence of a partition, but lost consistency. I'd rather sacrifice availability. Or only let people commit to one of the repos. I am worried that Apache INFRA don't realize that a distributed transactional system is a difficult beast. Julian On Wed, May 22, 2019 at 2:17 PM Kevin Risden wrote: > > I think there have been some delays syncing between github and gitbox. I > think it depends which origin you are pointing at and which commit is there. > > Since both github and gitbox are masters, we can actually push to either > one and I would guess there is some checks to make sure commits don't have > issues but a pull might be slightly out of date? > > Kevin Risden > > > On Wed, May 22, 2019 at 5:12 PM Julian Hyde wrote: > > > I pushed a commit to the master branch this morning and it was > > rejected due to an intervening commit from Laurent. But that > > intervening commit did not appear when I did 'git fetch origin'. > > > > Just now, I was just able to push successfully. In both cases I used > > 'git push origin master' from the command line, without the '-f' > > (force) flag. > > > > You can see that both commits generated emails: [1] [3]. Laurent's > > intervening commit also generated an email: [2]. > > > > All, Has anyone else seen this behavior? > > > > Laurent, What command did you use to commit? > > > > Julian > > > > [1] > > https://lists.apache.org/thread.html/ac70079b65305a9040bfe2ed0720768343d2f6dc038b9a63c081ff0b@%3Ccommits.calcite.apache.org%3E > > > > [2] > > https://lists.apache.org/thread.html/bddfd2ad8cbe3148820b334b1554fae23d4e05a2b00967b71e5d7191@%3Ccommits.calcite.apache.org%3E > > > > [3] > > https://lists.apache.org/thread.html/335c042be3fbd042f327e9742d45de790d731d8a7f883294dacf9322@%3Ccommits.calcite.apache.org%3E > >
Re: Rejected commits to master
I think there have been some delays syncing between github and gitbox. I think it depends which origin you are pointing at and which commit is there. Since both github and gitbox are masters, we can actually push to either one and I would guess there is some checks to make sure commits don't have issues but a pull might be slightly out of date? Kevin Risden On Wed, May 22, 2019 at 5:12 PM Julian Hyde wrote: > I pushed a commit to the master branch this morning and it was > rejected due to an intervening commit from Laurent. But that > intervening commit did not appear when I did 'git fetch origin'. > > Just now, I was just able to push successfully. In both cases I used > 'git push origin master' from the command line, without the '-f' > (force) flag. > > You can see that both commits generated emails: [1] [3]. Laurent's > intervening commit also generated an email: [2]. > > All, Has anyone else seen this behavior? > > Laurent, What command did you use to commit? > > Julian > > [1] > https://lists.apache.org/thread.html/ac70079b65305a9040bfe2ed0720768343d2f6dc038b9a63c081ff0b@%3Ccommits.calcite.apache.org%3E > > [2] > https://lists.apache.org/thread.html/bddfd2ad8cbe3148820b334b1554fae23d4e05a2b00967b71e5d7191@%3Ccommits.calcite.apache.org%3E > > [3] > https://lists.apache.org/thread.html/335c042be3fbd042f327e9742d45de790d731d8a7f883294dacf9322@%3Ccommits.calcite.apache.org%3E >
Rejected commits to master
I pushed a commit to the master branch this morning and it was rejected due to an intervening commit from Laurent. But that intervening commit did not appear when I did 'git fetch origin'. Just now, I was just able to push successfully. In both cases I used 'git push origin master' from the command line, without the '-f' (force) flag. You can see that both commits generated emails: [1] [3]. Laurent's intervening commit also generated an email: [2]. All, Has anyone else seen this behavior? Laurent, What command did you use to commit? Julian [1] https://lists.apache.org/thread.html/ac70079b65305a9040bfe2ed0720768343d2f6dc038b9a63c081ff0b@%3Ccommits.calcite.apache.org%3E [2] https://lists.apache.org/thread.html/bddfd2ad8cbe3148820b334b1554fae23d4e05a2b00967b71e5d7191@%3Ccommits.calcite.apache.org%3E [3] https://lists.apache.org/thread.html/335c042be3fbd042f327e9742d45de790d731d8a7f883294dacf9322@%3Ccommits.calcite.apache.org%3E
