On 4 December 2017 at 18:19, Dan Rue dan.rue@linaro.org wrote:
Greg pushed stable rc today. Several times.
4.4.104-rc was pushed 3 times. The first should be ignored completely as it had fewer commits than the subsequent two pushes, which were the same except for the commit timestamps: 4.4 rc build 58: beab94f6167d4bd198b688089fc68da0ff25ee0c (incomplete) 4.4 rc build 59: e1e2e6c522e173ad7b0a1261c7a9954b395c6a8b (same as 60) 4.4 rc build 60: 998ef014deff90a607c74c706b5bed3115931b61 (same as 59)
4.9.67-rc had two duplicate pushes (except for commit timestamps). I was able to cancel some of the lava jobs for build 104. 4.9 rc build 104: a89dba36bba844368f48cb55538656e5ff75354d (cancelled) 4.9 rc build 105: 3efaa252aaafecdd666059be175f88ebd95ea236 (same as 104)
4.14.4 was pushed 3 times. The first should be ignored completely as it had fewer commits than the subsequent two pushes, which were the same except for the commit timestamps: 4.14 rc build 14: 062b4a676f56ea414dabae6cf34c5e0f57de3931 (incomplete) 4.14 rc build 15: c2bf04f2ec6277782cbecd0c98a3c2a1a306dc16 (same as 16) 4.14 rc build 16: 95aa1a118d82e935ec7065345cbaff945d4100bf (same as 15)
The build numbers referenced are the kernel trigger numbers, which may or may not be the same as the actual per-board build numbers that show up in the lava job names. The only jobs I cancelled are the 4.9 build. The rest I will let run. I'll reply to the incomplete branch results and say as much, as they come through.
When we report results, we (always) need to verify the kernel revs match what is the current branch, since we are going to have multiple results.
The canceled jobs might show up as a baseline in qa-reports. I'll check whether this is actually the case. If that happens the diff in results will be calculated incorrectly. In such case the status in qa-reports (admin) should be marked as 'not finished' (checkbox unchecked). This way qa-reports will ignore the results and take previous 'finished' set.
milosz