I'm doing some pinning in ci to see if we can nail down which commits caused performance regressions, and to confirm that there wasn't also a tht change that contributed. Here are the results so far:
The commit we reverted, to confirm that the full regression was present after this merged:
f94d76cb322f14a001e7990c0918a67f1a09ea16 ControllerServiceChain 722.0
It was. Another commit about halfway between that one and the commit used for the promotion prior to the regression showing up:
b5359cedf10dab59a11effb0ebed383ad23c73a6 ControllerServiceChain 169.0
Better, but still not back to the 30 seconds or so that it used to be. I've pushed two new pins starting at https://review.openstack.org/#/c/460188 to test another halfway point between the last known good, and the last known good itself to verify it still yields the same results it did before the promotion.
I'm doing some pinning in ci to see if we can nail down which commits caused performance regressions, and to confirm that there wasn't also a tht change that contributed. Here are the results so far:
The commit we reverted, to confirm that the full regression was present after this merged: 001e7990c0918a6 7f1a09ea16 ControllerServi ceChain 722.0
f94d76cb322f14a
It was. Another commit about halfway between that one and the commit used for the promotion prior to the regression showing up: 9a11effb0ebed38 3ad23c73a6 ControllerServi ceChain 169.0
b5359cedf10dab5
Better, but still not back to the 30 seconds or so that it used to be. I've pushed two new pins starting at https:/ /review. openstack. org/#/c/ 460188 to test another halfway point between the last known good, and the last known good itself to verify it still yields the same results it did before the promotion.