On Thu, Jul 02, 2015 at 06:50:29PM +0300, Oleg Ponomarev wrote:
> The bug is using incorrect location score during failover and replace
> action. Thus the idea of the test is to introduce a situation in which
> it's profitably to perform failover and replace move in order to
> optimize cluster score (if ignoring location tags). Considering location
> tags another final placement is optimal.
> 
> In this test the optimal placement should locate primary instances to
> nodes 1 and 3 with secondaries on the second node.
> 
> Signed-off-by: Oleg Ponomarev <[email protected]>
> ---
>  Makefile.am                           |  1 +
>  test/data/htools/hbal-location-2.data | 10 ++++++++++
>  test/hs/shelltests/htools-hbal.test   |  9 ++++++++-
>  3 files changed, 19 insertions(+), 1 deletion(-)
>  create mode 100644 test/data/htools/hbal-location-2.data

LGTM.

However, while it is good practise for first write tests and see them fail,
we usually organise our patch series in such a way that after each patch
all tests pass. In this way, binary searches, e.g., by git bisect, are
not disturbed by spurious test failures.

So I will push your patches in the other order, with the fix first.

-- 
Klaus Aehlig
Google Germany GmbH, Dienerstr. 12, 80331 Muenchen
Registergericht und -nummer: Hamburg, HRB 86891
Sitz der Gesellschaft: Hamburg
Geschaeftsfuehrer: Graham Law, Christine Elizabeth Flores

Reply via email to