Alexey Serbin has posted comments on this change. ( http://gerrit.cloudera.org:8080/15274 )
Change subject: [clock] relax version constraint for NTP response ...................................................................... Patch Set 1: (5 comments) http://gerrit.cloudera.org:8080/#/c/15274/1//COMMIT_MSG Commit Message: http://gerrit.cloudera.org:8080/#/c/15274/1//COMMIT_MSG@13 PS1, Line 13: interestingly enough, the dedicated NTP servers of : other EC2 t2.large instances in my test cluster behaved as expected > I'd expect the NTP server behavior to be a function of the NTP server softw Nope, there was not a difference: all are ami-0198ddd64383b5bb0 (AMI named 'cloudera-systest-base-centos-7.5'). The only difference is the type of EC2 instances: in my test cluster, I have four t2.large instances and one r5ad.xlarge instance. All t2.large instances show the same conformant behavior, and only the r5ad.xlarge instance behaves in a different way. All NTP servers report they are of version 4, but there is a difference in behavior. In addition, all t2.large instances report precision of 0.000000015 seconds while r5ad.xlarge report precision of 0.000003815 seconds. Also, TX timestamping is different. In other words, apparently those are different NTP server implementations. `chronyc ntpdata 169.254.169.123` for t2.large: Remote address : 169.254.169.123 (A9FEA97B) Remote port : 123 Local address : 10.65.11.29 (0A410B1D) Leap status : Normal Version : 4 Mode : Server Stratum : 3 Poll interval : 8 (256 seconds) Precision : -26 (0.000000015 seconds) Root delay : 0.000259 seconds Root dispersion : 0.000275 seconds Reference ID : 0A81F8C4 () Reference time : Mon Feb 24 06:09:52 2020 Offset : +0.000034901 seconds Peer delay : 0.000251005 seconds Peer dispersion : 0.000000373 seconds Response time : 0.000113064 seconds Jitter asymmetry: +0.00 NTP tests : 111 111 1111 Interleaved : No Authenticated : No TX timestamping : Daemon RX timestamping : Kernel Total TX : 902 Total RX : 902 Total valid RX : 902 `chronyc ntpdata 169.254.169.123` for r5ad.xlarge: Remote address : 169.254.169.123 (A9FEA97B) Remote port : 123 Local address : 10.65.14.145 (0A410E91) Leap status : Normal Version : 4 Mode : Server Stratum : 3 Poll interval : 6 (64 seconds) Precision : -18 (0.000003815 seconds) Root delay : 0.000183 seconds Root dispersion : 0.000259 seconds Reference ID : A9FEA97A () Reference time : Mon Feb 24 06:12:20 2020 Offset : +0.000004004 seconds Peer delay : 0.000117991 seconds Peer dispersion : 0.000003855 seconds Response time : 0.000014430 seconds Jitter asymmetry: -0.39 NTP tests : 111 111 1111 Interleaved : No Authenticated : No TX timestamping : Kernel RX timestamping : Kernel Total TX : 1262 Total RX : 1262 Total valid RX : 1262 http://gerrit.cloudera.org:8080/#/c/15274/1//COMMIT_MSG@17 PS1, Line 17: Secion > Section Done http://gerrit.cloudera.org:8080/#/c/15274/1//COMMIT_MSG@20 PS1, Line 20: the > of the Done http://gerrit.cloudera.org:8080/#/c/15274/1//COMMIT_MSG@29 PS1, Line 29: As a side note, : I also explored options to implement a test scenario using MiniChronyd, : but I found no way to configure chronyd to behave in a non-conforming : manner described above. > Seems like it'd be good to have the ability to create a "mock" chronyd that Sure, that's my concern as well. I'm planning to take a closer look at the idea of a mock NTP server this week. http://gerrit.cloudera.org:8080/#/c/15274/1/src/kudu/clock/builtin_ntp.cc File src/kudu/clock/builtin_ntp.cc: http://gerrit.cloudera.org:8080/#/c/15274/1/src/kudu/clock/builtin_ntp.cc@732 PS1, Line 732: Secion > Section Done -- To view, visit http://gerrit.cloudera.org:8080/15274 To unsubscribe, visit http://gerrit.cloudera.org:8080/settings Gerrit-Project: kudu Gerrit-Branch: master Gerrit-MessageType: comment Gerrit-Change-Id: I1df3549616bbced696380458978c33264ed636c9 Gerrit-Change-Number: 15274 Gerrit-PatchSet: 1 Gerrit-Owner: Alexey Serbin <[email protected]> Gerrit-Reviewer: Adar Dembo <[email protected]> Gerrit-Reviewer: Alexey Serbin <[email protected]> Gerrit-Reviewer: Kudu Jenkins (120) Gerrit-Comment-Date: Mon, 24 Feb 2020 17:16:04 +0000 Gerrit-HasComments: Yes
