While some things are up in the air we should step this forward as good as we can, so @ubuntu-mir Team pleas ack and set it so the next status (if ok) so that the security Team can do an official check and ack as well.
** Description changed: - Note: I know it is the template so far, but after the discussions at the - sprint I want something we can start working on together. - - Background: after evaluation it was considered easier to maintain to - provide a good and secure ntp experience as well as some people asking - me if it could be preferred. - --- MIR --- 1. Availability: The package is Ubuntu universe and builds for the architectures it is designed to work on. - 2. Rationale: NTP in general is needed quite a lot, but we want to - exchange ntpd which is the current implementation in main with chrony - for 18.04. + 2. Rationale: + 2.1 NTP in general is needed quite a lot, but we want to exchange ntpd + which is the current implementation in main with chrony for 18.04. + 2.2 Security: chrony was considered easier to be maintained easier in + terms of security and provide a more modern ntp experience as well. + 2.3 Efficiency: Furthermore several cloud people seem to be interested to + change to chrony in the guests for its lower memoy/cpu footprint + (efficiency I guess). + 2.4 related to this MIR 6 years ago this is the same but for Fedora. + See: https://fedoraproject.org/wiki/Features/ChronyDefaultNTP + IIRC some limitations that were present have been eliminated since, so + it is even better than it was back then. + 2.5 In general one has to realize that in a systemd-timesync world + ntp/chrony are mostly for the "serving" portion of an ntp service, and + not so much about the client (unless you the better accuracy vs + timesyncd is needed). 3. Security: In fact the request came in by security Team, so I guess I call this section done -- EVERYTHING BELOW TBD FOR NOW -- 3. Quality assurance - 3.1 configuration ease - works after installation - 3.2 no high prio debconf - 3.3 usability (no major issues in Debian nor Ubuntu) - asked Paul in regard to the ntp charm in comment #5 - 3.4 long-term >=high bugs (none in Debian nor Ubuntu) - 3.5 Debian/Ubuntu bugs look reasonable maintained - 3.6 does not deal with hard to support exotic hardware (other than ntpd - btw). If used this can be done through universe package GPSD (no - dependency) - 3.7 Test suite runs on build (some skipped if not env applicable) - 3.8 debian/watch exists - 3.9 not depending on obsoleted packages - + 3.1 configuration ease - works after installation + 3.2 no high prio debconf + 3.3 usability (no major issues in Debian nor Ubuntu) + asked Paul in regard to the ntp charm in comment #5 + 3.4 long-term >=high bugs (none in Debian nor Ubuntu) + 3.5 Debian/Ubuntu bugs look reasonable maintained + 3.6 does not deal with hard to support exotic hardware (other than ntpd + btw). If used this can be done through universe package GPSD (no + dependency) + 3.7 Test suite runs on build (some skipped if not env applicable) + 3.8 debian/watch exists + 3.9 not depending on obsoleted packages 4.1 It does not face graphical UI 4.2 It is unfortunately not internationalized as far as I could see in the source 5. Dependencies - there is one not in main libtomcrypt - We don't want it in main either, instead we want to fix bug 1744328 and then use libnss which is in main already. + We don't want it in main either, instead we want to fix bug 1744328 and then use libnss which is in main already. 6. Not found major Policy or FSH violations that would have to be fixed. 7. Maintenance - 7.1 Upstream - is maintained well (and better than ntpd it seems - according to some discussisons) - 7.2 Ubuntu - Owning Team would be Ubuntu (in exchange to drop ntp) + 7.1 Upstream - is maintained well (and better than ntpd it seems + according to some discussisons) + 7.2 Ubuntu - Owning Team would be Ubuntu (in exchange to drop ntp) 8. Background information: - Fulfills the same role as ntp, yet according to the security Team would - be preferred for them. + Fulfills the same role as ntp, yet according to the security Team would + be preferred for them. --- Affected Packages --- I'll add all those as bug tasks. Once the MIR has passed the state of uncertainty (e.g. would it be blocked by one of the dependent bug being not doable at all) then please work on these into 18.04. Here a list what is affected in the listed packages: Maas - needs to change dependencies and maybe template cloud-init - needs to support writing ntp config to chrony instead of ntpd ceph-base - change recommends from ntpd to chrony (it only intends to get good time and doesn't care via which dameon that is, so that should be ok to be change) ntp charm - switch to chrony for >=18.04 chrony - MIR itself (discussion here and eventually seeding) --- Depending on further Bugs --- In my initial evaluation I uncovered (and filed) a set of bugs that I consider requirement to make it fully ready: Reminder - tracking state here might be out of sync, I'll only change them to Done once complete and not care about interim status changes. OPEN - bug 1718227 - convert ifup hooks to networkd compatible hooks OPEN - bug 1744662 - add chrony apparmor profile OPEN - bug 1744328 - make src:libnss libfreebl3 usable by other programs OPEN - bug 1744664 - use Ubuntu time servers ** Also affects: maas (Ubuntu) Importance: Undecided Status: New ** Also affects: cloud-init (Ubuntu) Importance: Undecided Status: New ** Also affects: ceph (Ubuntu) Importance: Undecided Status: New ** Also affects: ntp-charm Importance: Undecided Status: New -- You received this bug notification because you are a member of Ubuntu Bugs, which is subscribed to Ubuntu. https://bugs.launchpad.net/bugs/1744072 Title: MIR Chrony in 18.04 To manage notifications about this bug go to: https://bugs.launchpad.net/ntp-charm/+bug/1744072/+subscriptions -- ubuntu-bugs mailing list ubuntu-bugs@lists.ubuntu.com https://lists.ubuntu.com/mailman/listinfo/ubuntu-bugs
