)]}'
{"/COMMIT_MSG":[{"author":{"_account_id":2813,"name":"Mehdi Abaakouk","email":"sileht@sileht.net","username":"sileht"},"change_message_id":"da0199fed4bdb057522727a58a6c298f0df2648b","unresolved":false,"context_lines":[{"line_number":4,"context_line":"Commit:     Su Zhengwei \u003csu.zhengwei@zte.com.cn\u003e"},{"line_number":5,"context_line":"CommitDate: 2016-09-12 08:23:23 +0000"},{"line_number":6,"context_line":""},{"line_number":7,"context_line":"rate_of_change transformer output inconsistent with original samples"},{"line_number":8,"context_line":""},{"line_number":9,"context_line":"The meter_source poll the network-meters and keep them in cache,"},{"line_number":10,"context_line":"then network_source use the same data from the cache to generate"}],"source_content_type":"text/x-gerrit-commit-message","patch_set":4,"id":"9a89bdaa_9c41580e","line":7,"updated":"2016-09-12 08:45:23.000000000","message":"Do we really want to have everything during a polling with the same timestamp, when you can have when the sample have really been computed ?\n\nIf the transformer timestamp is 1 seconds after this is just because the sample was computed 1 second later, I think is OK.\n\nAlso the ceilometerclient output removes microseconds \nIf the time between two polling was \"exactly\" 10 minutes, then yes you should got 41.3, but I wonder if reality is so precise. Perhaps you got 41.231281198  because the exact time between both polling was 9.98336106489104116222, is that possible ?\n\nHave you checks the real timestamp in the database ?","commit_id":"496d0e77bb75c8342d5629adf38a4ed941f318b7"},{"author":{"_account_id":2813,"name":"Mehdi Abaakouk","email":"sileht@sileht.net","username":"sileht"},"change_message_id":"da0199fed4bdb057522727a58a6c298f0df2648b","unresolved":false,"context_lines":[{"line_number":13,"context_line":"Same problem between meter_source and disk_source, if they use"},{"line_number":14,"context_line":"the same polling interval. That makes me confused."},{"line_number":15,"context_line":""},{"line_number":16,"context_line":"So it\u0027s better to use the same timestamp during one polling."},{"line_number":17,"context_line":""},{"line_number":18,"context_line":"Closes-bug: #1620947"},{"line_number":19,"context_line":""}],"source_content_type":"text/x-gerrit-commit-message","patch_set":4,"id":"9a89bdaa_dc968093","line":16,"updated":"2016-09-12 08:45:23.000000000","message":"I\u0027m not sure.","commit_id":"496d0e77bb75c8342d5629adf38a4ed941f318b7"}],"ceilometer/agent/manager.py":[{"author":{"_account_id":2813,"name":"Mehdi Abaakouk","email":"sileht@sileht.net","username":"sileht"},"change_message_id":"da0199fed4bdb057522727a58a6c298f0df2648b","unresolved":false,"context_lines":[{"line_number":153,"context_line":"        cache \u003d {}"},{"line_number":154,"context_line":"        discovery_cache \u003d {}"},{"line_number":155,"context_line":"        poll_history \u003d {}"},{"line_number":156,"context_line":"        polling_timestamp \u003d timeutils.utcnow().isoformat()"},{"line_number":157,"context_line":"        for source_name in self.pollster_matches:"},{"line_number":158,"context_line":"            for pollster in self.pollster_matches[source_name]:"},{"line_number":159,"context_line":"                key \u003d Resources.key(source_name, pollster)"}],"source_content_type":"text/x-python","patch_set":4,"id":"9a89bdaa_5c827053","line":156,"updated":"2016-09-12 08:45:23.000000000","message":"This change the meaning, now polling_timestamp is \u0027the polling start timestamp\u0027 and no more when the data is really polled.\n\nI don\u0027t known well the Ceilometer part, but this change looks suspicious.","commit_id":"496d0e77bb75c8342d5629adf38a4ed941f318b7"},{"author":{"_account_id":17289,"name":"Su Zhengwei","email":"su.zhengwei@zte.com.cn","username":"su-zhengwei"},"change_message_id":"a305b37469e27bda16210e2f8ea7f53d0e48636d","unresolved":false,"context_lines":[{"line_number":153,"context_line":"        cache \u003d {}"},{"line_number":154,"context_line":"        discovery_cache \u003d {}"},{"line_number":155,"context_line":"        poll_history \u003d {}"},{"line_number":156,"context_line":"        polling_timestamp \u003d timeutils.utcnow().isoformat()"},{"line_number":157,"context_line":"        for source_name in self.pollster_matches:"},{"line_number":158,"context_line":"            for pollster in self.pollster_matches[source_name]:"},{"line_number":159,"context_line":"                key \u003d Resources.key(source_name, pollster)"}],"source_content_type":"text/x-python","patch_set":4,"id":"9a89bdaa_f2c9ad7a","line":156,"in_reply_to":"9a89bdaa_5c827053","updated":"2016-09-12 09:04:48.000000000","message":"The timestamp does\u0027t represent the time when the data is really polled, because some samples maybe produced by data from cache.\nSame data, same timestamp.If we used the same cache why did\u0027t we use the same timestamp?","commit_id":"496d0e77bb75c8342d5629adf38a4ed941f318b7"}]}