[07:12] <cpaelzer> xnox: did my qemu upload miss nettle/hogweed by 2 hours :-/
[07:13] <cpaelzer> I was confused seeing "Build for superseded Source" after the 12h risc build job completed, but then found your ubuntu2 uploaded just slightly after it
[07:14]  * cpaelzer exuses at the builders for twice the load to to this 2h miss - or if it wasn't for the 2h stares at xnox for not checking the build time of ubuntu1 :-P
[07:16] <cpaelzer> hmm was even 6h TBH, so yeah mabye just unlucky timing and4 more hours to go for the new risc build
[07:52] <seb128> I'm curious, why was that nettle rebuild for?
[07:52] <seb128> '  * No change rebuild against new libnettle8 and libhogweed6 ABI.'
[07:52] <seb128> but those are binaries from the same source package
[08:07] <seb128> xnox, would it make sense to have a transition tracker set up for nettle?
[08:50] <sil2100> I'll be looking at opensaml autohint right now
[09:34] <xnox> seb128:  transition tracker => yes
[09:34] <xnox> cpaelzer:  i looked at -release pocket to generate 40+ uploads =/ some of them may have been redundant.
[09:34] <xnox> (i.e. if they were _building_ in -proposed)
[09:34] <xnox> sorry
[09:35] <seb128> xnox, could you set it up?
[09:38] <xnox> seb128:  yes, doing
[09:39] <seb128> xnox, thanks
[09:39] <seb128> xnox, isn't the auto-tracker thing supposed to do that automatically?
[09:39] <xnox> seb128:  auto-tracker is dimitri running rsync on dists, running the script, committing the trackers.....
[09:39] <xnox> which is what dimitri is doing
[09:39] <seb128> I see
[09:40] <seb128> we should set a proper job so less work for Dimitri :)
[09:40] <xnox> seb128:  i haven't worked out how ben is deployed on snakefruit, to automate =)
[09:56] <Laney> xnox: we should get lxd installed there, and then we can try to get a modern ben in a tractable way
[09:56] <Laney> vorlon / sil2100: opinions?
[11:04]  * xnox pushes the autotrackers to the right branch....
[12:11] <kanashiro> mwhudson: I am considering to upload the cadvisor debdiff attached in LP #1884663 at the end of the day (disabling containerd support) to unblock stuff in proposed, if you have any objection please let me know
[12:11] <ahasenack> good morning
[12:21] <kanashiro> today I am going to keep working on ruby packages, a bunch of them are stuck in proposed
[12:44] <eoli3n> zyga you're nfs detection has been merged into snapd, when will it be packaged for ubuntu ?
[12:44] <eoli3n> your
[12:44] <rafaeldtinoco> morning o/
[12:51] <juliank> xnox, rbalint, ddstreet: It seems ddstreet dropped ondemand.service in systemd, causing system to boot in performance mode now, which means it runs in turbo all the time
[12:52] <xnox> maybe we can resurrect it, fix it up for pstates and still drop down to something more sensible instead of performance?
[12:52] <juliank> xnox: Well it works for pstates
[12:53] <xnox> juliank:  note Wimpress was also asking for GUI toggles to do performance or not.
[12:53] <juliank> xnox: It checks which governors are available, and then sets the best one
[12:53] <juliank> xnox: And it only understands interactive,ondemand,powersave, and powersave is the only overlap with pstates, and the recommended governor
[12:54] <xnox> imho we should drop interactive,ondemand, and just do performance or powersave, and do powersave by default after a delay?
[12:54] <juliank> xnox: I also think this should not be static, e.g. we should always use powersave governor, but its preference ! should be balanced_performance on AC and balanced_power on BAT
[12:54] <juliank> for pstates that is
[12:55] <xnox> juliank:  so udev, when detecting ac/battery change should start governor@balanced_power / blaanced_performance?
[12:55] <xnox> unit, or just write to the file
[12:56] <juliank> xnox: Well maybe, or thermald should set it, or upower
[12:56]  * xnox is not sure what is monitoring and reacting to ac/battery
[12:56] <juliank> xnox: like I have tlp installed to do that stuff
[12:56]  * xnox thought upower is dead?
[12:56] <juliank> xnox: is it?
[12:56] <juliank> it's still used!
[12:57] <xnox> it's in main, and seeded, so i guess upower is being used =)
[12:57] <xnox> is balanced_power the right thing to do on servers too?
[12:57] <juliank> Settings should give you options to configure the governor (performance vs powersave, though, who really wants performance, it just keeps CPU at max turbo all the time)
[12:57] <juliank> And then energy_performance_available_preferences = default performance balance_performance balance_power power
[12:58] <juliank> I guess performance governor or powersave governor with performance preference is right for servers
[12:58] <juliank> though if you run a home server and want to save watts while idle, you might want balance_performance
[13:00] <xnox> balanced_power sounds right
[13:00] <xnox> as a better default than current
[13:01] <juliank> xnox: No I think you want balanced_performance as the default, balanced_power turns of turbo or something
[13:01] <xnox> juliank:  wanna open a bug report against systemd, and ask to resurrect ondemand.service, and make it do new shiny things?
[13:01] <xnox> i.e. "don't drop it, make it pick better thing on better hardware"
[13:02] <juliank> With balanced_power, I max out at 2.4 GHz instead of 3.4 GHz which is not super useful
[13:02] <juliank> (average across cores)
[13:02] <seb128> juliank, xnox, we have gamemode since focal which is basically a service and api to ask the governor to the set to performance (which some games use now)
[13:02] <juliank> I think other distributions set the default governor to powersave in kernel config, and hence don't need that systemd unit?
[13:03] <seb128> there is also a gnome-shell-extension-gamemode which gives you an UI toggle for it
[13:04] <juliank> seb128: I'd like this neatly integrated without being gaming specific really, like windows has the slide from performance - balanced perf - balanced power - power, we should have too
[13:04] <xnox> juliank:  can we change kernel config to do balanced_performance? (CC apw bjf) and win?
[13:04] <seb128> juliank, that's something we have been discussing doing next cycle in desktop
[13:04] <xnox> juliank:  yeah, Wimpress is asking for that.....
[13:04] <cpaelzer> LocutusOfBorg: hi, you merged devscripts 2.20.4 into groovy. But it now shows up in component mismatches
[13:04] <juliank> xnox: balanced_performance is the default for the powersave governor, but we boot with the performance governor to make initial boot faster (because CPU cores run at max turbo constantly until we are booted up)
[13:04] <cpaelzer> LocutusOfBorg: due to "Add pristine-tar to Recommends.  Closes: #961532"
[13:05] <xnox> juliank:  seb128: but a slider/ui/etc to change the default, doesn't mean we don't need to have a sensible default, or can't have it already?
[13:05] <juliank> xnox: I'm not sure this makes smuch of a difference
[13:05] <xnox> hm
[13:05] <seb128> xnox, juliank, sensible default always make sense
[13:05] <juliank> xnox: having performance vs powersave x balanced_performance
[13:05] <cpaelzer> LocutusOfBorg: would you mind uploading an ubuntu2 version which does revert this minor change, as otherwise this explodes into follow on dependencies at https://people.canonical.com/~ubuntu-archive/component-mismatches.svg
[13:05] <juliank> xnox: seb128: context was wrong, I was replying to my previous message, not to the latest xnox once about sensible default
[13:06] <xnox> juliank: well, we didn't have balanced_performance before. Have we measured if it is slower to boot, than performance=>then=>powersave?
[13:06] <juliank> xnox: seb128: I do agree we need a sensible default (powersave x balanced_performance)
[13:06] <juliank> xnox: Probably not?
[13:06] <xnox> do we need to ask cking to like redo all of those measurements again?
[13:07] <juliank> maybe?
[13:07] <xnox> also performance inside VMs kills the host basically
[13:38] <cpaelzer> kanashiro: you are on +1 duty this week right?
[13:39] <kanashiro> cpaelzer: yes
[13:39] <cpaelzer> kanashiro: I wanted to ask if I could "request" something from you that I've seen while prepping the MIR meeting
[13:39] <cpaelzer> kanashiro: I ahve above asked LocutusOfBorg already, but I'm unsure if he's seen it and is available
[13:40] <cpaelzer> kanashiro: the recent upload of src:devscripts brought in a debian cahnge to depend on pristine-tar
[13:40] <cpaelzer> kanashiro: that is a component mismatch which blocks things in groovy-proposed
[13:40] <cpaelzer> kanashiro: I wondered if you could set yourself a reminder and if LocutusOfBorg didn't get to it e.g. until tomorrow EOD then you might do that
[13:40] <cpaelzer> kanashiro: in case you are ok
[13:41] <rbalint> xnox, juliank, seb128, ddstreet please open a bug against systemd in LP, but i think we should not resurrect the service but let kernel and desktop sort governor out
[13:42] <rbalint> we can document the reasoning in the bug if we decide one way or an other
[13:42] <juliank> rbalint: ack
[13:42] <kanashiro> cpaelzer: ok, I can take a look at devscripts
[13:45] <seb128> rbalint, xnox, juliank, if we really default to performance now that seems like an important regression and worth fixing now while we figure out the right way to do that if we prefer to change
[13:46] <juliank> yes
[13:46] <rbalint> there is quite a bit of history here: https://git.launchpad.net/~ubuntu-core-dev/ubuntu/+source/systemd/commit/?id=65f46a7d14b335e5743350dbbc5b5ef1e72826f7
[13:47] <rbalint> seb128, server and cloud is the highest priority for Ubuntu and ubuntu-desktop can pull in something to override the governor
[13:47] <rbalint> seb128, if you don't have desktop you are probably not sitting next to the system, so noise is not an issue :-)
[13:48] <seb128> rbalint, earth thanks human for such way of thinking right?
[13:50] <rbalint> seb128, earth won't thank you if you buy one more server because 10 was not fast enough either
[13:51] <rbalint> seb128, please check the bugs referenced in ddstreet's commit and possibly the new one and make your point in the new one
[13:56] <rbalint> seb128, we should flip again only after we carefully explored the topic (again) and found that flipping is the right thing to do
[13:57] <seb128> rbalint, sorry, we had our meeting at the same time
[13:59] <seb128> rbalint, ddstreet, Wimpress I guess my main issue there is that the change was neither discussed/announced on devel or discourse nor anyone did reach out to other team to tell them they need to update their product if they wanted to keep the behaviour
[14:00] <seb128> that's not how changes with a such impact should be handled
[14:03] <rbalint> seb128, the change is only in devel (groovy) and arrived with a merge from Debian, we are well in time to handle it, no panic is needed
[14:04] <seb128> rbalint, I'm not panicing, don't worry :)
[14:05] <rbalint> seb128, on my scale, you do :-)
[14:06] <seb128> rbalint, sorry, I just don't like having those topic coming just because someone did notice and raised it there, ideally such changes should be announced before upload / at upload time and discussed with the impacted flavors
[14:07] <seb128> (you can tell me that was about to happen but I somewhat doubt it was)
[14:08] <juliank> rbalint: I don't think this has been analysed carefully
[14:08] <juliank> rbalint: Most of the bug comments are unrelated musings about pstate vs other solutions
[14:10] <rbalint> juliank, ok, please open the LP bug and have a careful analysis there because only very few people look at irc logs
[14:11] <juliank> rbalint: Like why is the bug 1806012 discussing pstate vs other drivers, when the bug report was about toggling the pstate governor from powersave to performance, it does not make much semse
[14:17] <rbalint> juliank, i don't know, i have not touched that bug, but i'll be happy to participate in the discussion in a new one
[14:18] <rbalint> i think the default is better to be controlled by kernel, it can be ondemand and custom cloud kernels can switch it to performance
[14:19] <sil2100> Looking at the migration of orcania and yder
[14:19] <rbalint> but more on it in the bug, we have enough time to resolve situation in this cycle
[14:21] <kanashiro> cpaelzer: re devscripts: version 2.20.4ubuntu1 migrated to the release pocket hours ago. At the moment, devscripts recommends pristine-tar, you mean we should downgrade it to suggests?
[14:23] <juliank> rbalint: seb128 I reported bug 1885730 to track the ondemand stuff, at least the pstate stuff, I also subscribed desktop and server so both sides (so to speak) can give feedback
[14:24] <seb128> juliank, thanks
[14:24] <juliank> I also add a linux task for kernel team so we can look at why our default is performance in the kernel
[14:24] <rbalint> juliank, thanks, i was about to do that, too
[14:26] <cpaelzer> kanashiro: it is a current component mismatch
[14:26] <cpaelzer> kanashiro: and I don't see anyone pulling in all the dependencies of it - see https://people.canonical.com/~ubuntu-archive/component-mismatches.svg
[14:36] <bdmurray> xnox: I'm trying to sort out the ubuntu-release-upgrader autopkgtest failure and could use some help. https://objectstorage.prodstack4-5.canonical.com/v1/AUTH_77e2ada1e7a84929a74ba3b87153c0ac/autopkgtest-groovy/groovy/amd64/u/ubuntu-release-upgrader/20200625_034545_a8ed9@/log.gz
[14:36] <bdmurray> the call to apt-key verify for the xenial dist-upgrader is failing
[15:14] <xnox> bdmurray:  groovy has removed 2012 key, and only trusts 2018 key. To verity xenial, from groovy, one may not use "just" apt-key with trusted.gpg/trusted.gpg.d keyrings, one has to use /usr/share/keyrings/ubuntu-archive-keyring.gpg
[15:14] <xnox> bdmurray:  why do you care to verify that groovy can "dist-upgrade?" to xenial?
[15:14] <xnox> bdmurray:  this is similar to when we removed the old 2004/2008 keys, in xenial. and then stopped dual signing.
[15:18] <bdmurray> xnox: its testing an upgrade from trusty to xenial and I'm still curious why its broken and setting apt-key "Dir::Etc::Trusted" doesn't seem to work
[15:19] <xnox> bdmurray:  ubuntu-keyring in groove _removed_ _all_ keys from trusted.gpg* that are used to sign xenial.
[15:19] <xnox> bdmurray:  to test upgrade from trusty to xenial, you must have trusty's trusted.gpg* somewhere, and use those for the upgrade. I.e. by creating a small apt DIR with the right files.
[15:20] <bdmurray> xnox: I get that and have added an old key with which I can manually verify using 'apt-file --key'
[15:20] <xnox> bdmurray:  which key did you add, and where?
[15:20] <xnox> where is the source code?
[15:21] <xnox> there is $ apt-config dump | grep trust
[15:21] <xnox> Dir::Etc::trusted "trusted.gpg";
[15:21] <xnox> Dir::Etc::trustedparts "trusted.gpg.d";
[15:21] <xnox> which are relative Dir::Etc "etc/apt";
[15:21] <xnox> which is relative Dir "/";
[15:21] <bdmurray>         #apt_pkg.config.set("Dir::Etc::Trusted",
[15:21] <bdmurray>         #                   "%s/ubuntu-keyring-2012-archive.gpg" % self.testdir)
[15:22] <xnox> normally, one creates custom Dir, and have things inside it.
[15:22] <bdmurray> right the above didn't work
[15:22] <xnox> bdmurray:  but what does Dir::Etc and Dir set to? cause it might be looking up /etc/apt/testdirname-inside-ubuntu-release-upgrader/ubuntu-keyring-2012-archive.gpg
[15:23] <xnox> bdmurray:  it's best to create "trusty-apt-dir" with like etc/apt/trusted.gpg.d/ubuntu-keyring-2012-archive.gpg
[15:23] <xnox> bdmurray:  nad set Dir option to absolute/path/to/trusty-apt-dir
[15:24] <bdmurray> '/tmp/autopkgtest.GBtKE7/build.SIJ/src/tests/data-sources-list-test//ubuntu-keyring-2012-archive.gpg'
[15:24] <xnox> bdmurray:  apt-config dump is useful to figure out what everything is set to.
[15:24] <xnox> bdmurray:  close enough, but ti's relative the build dir, which is not relative the source dir. thus i guess that filepath does not exist?!
[15:24] <bdmurray> that's from 'apt_pkg.config.get("Dir::Etc::Trusted") right before the verify call
[15:25] <bdmurray> I'm in an autopkgtest --shell-fail environment
[15:25] <xnox> ok
[15:25] <xnox> and where is ubuntu-keyring-2012-archive.gpg ?
[15:25] <xnox> can you do find/mlocate/locate/whatnot to find it?
[15:26] <xnox> cause => /tmp/autopkgtest.GBtKE7/build.SIJ/ is that ADTMPT? or where the sources are unpacked? or like were copied to?
[15:26] <bdmurray> ubuntu@autopkgtest:~$ file /tmp/autopkgtest.GBtKE7/build.SIJ/src/tests/data-sources-list-test//ubuntu-keyring-2012-archive.gpg
[15:26] <bdmurray> /tmp/autopkgtest.GBtKE7/build.SIJ/src/tests/data-sources-list-test//ubuntu-keyring-2012-archive.gpg: PGP/GPG key public ring (v4) created Fri
[15:26] <bdmurray>  May 11 14:15:36 2012 RSA (Encrypt or Sign) 4096 bits MPI=0xdf88bd4a3451d3e0...
[15:26] <bdmurray> its right where it belongs
[15:26] <xnox> that's good.
[15:28] <xnox> $ gpg --no-default-keyring --keyring /usr/share/keyrings/ubuntu-archive-keyring.gpg  --verify ./xenial.tar.gz.gpg ./xenial.tar.gz
[15:28] <xnox> gpg: Signature made Sat 16 Apr 2016 17:33:59 BST
[15:28] <xnox> gpg:                using DSA key 40976EAF437D05B5
[15:29] <xnox> gpg: Can't check signature: No public key
[15:29]  * xnox ponders what the xenial upgrade tarballs are signed with
[15:29] <bdmurray> xnox: use the the one from -updates
[15:29] <xnox> oh, good point
[15:29] <bdmurray> I know because I ran into that too. ;-)
[15:30] <xnox> and what does the test use? release or updates?
[15:30] <bdmurray> updates
[15:30] <bdmurray> because that is what is in all the meta-release files
[15:30] <xnox> bdmurray:  is the xenial.tar.gz somewhere?
[15:30] <xnox> or it got removed?
[15:30] <bdmurray> http://archive.ubuntu.com/ubuntu/dists/xenial-updates/main/dist-upgrader-all/current/
[15:30] <xnox> i mean inside autopkgtest --fail-shell
[15:30] <bdmurray> They have the same name but different content
[15:30] <xnox> / test run
[15:31] <bdmurray> its gets downloaded to a tempdir and removed
[15:31] <bdmurray> also this passes
[15:31] <bdmurray> "keyring = apt_pkg.config.get("Dir::Etc::Trusted")"
[15:35] <xnox> bdmurray:  btw, instead of copying a snippet, you can access old keys in /usr/share/keyrings
[15:38] <xnox> bdmurray:  https://paste.ubuntu.com/p/Z6VS47rbgV/
[15:38] <xnox> seems to work for me
[15:39] <xnox> bdmurray:  we used to ship keys in /etc/apt/trusted.gpg, then switched to /etc/apt/trusted.gpg.d frabments, and like pretty much deprecate apt-key usage directly.
[15:39] <xnox> bdmurray:  however /usr/share/keyrings/ubuntu-archive-keyring.gpg remains unchanged.
[15:39] <xnox> and has keys for the active series, and has the 2012 key even in groovy, because xenial is still alive.
[15:39] <xnox> so above code is correct on any series.
[15:40] <bdmurray> xnox: hmm okay, it still seems like something is wrong with setting "Dir::Etc::Trusted" but I'll submit a bug or something
[15:41] <vorlon> Laney: +1 on having lxd, +0 on "modern ben" :)
[15:42] <Laney> one which gets the levels right would be appreciated
[15:42] <vorlon> oh is there one of those now?
[15:42] <vorlon> then yes that interests me
[15:42] <Laney> but +1 on the former is enough for me to file a ticket
[15:42] <Laney> well, aiui yes
[15:43] <vorlon> (I still don't /like/ the transition tracker, but as long as others are going to use it, I'd rather it not mislead them)
[15:47] <bdmurray> xnox: Do you want me to upload that?
[15:49] <xnox> bdmurray:  yes please.
[15:49] <xnox> bdmurray:  i don't trust apt-key much =) and i have had pain using Dir:: options before too
[17:53] <bdmurray> rafaeldtinoco: regarding the SRU for bug 1680224, why is "d/p/lp1680224-fix-and-double-quoting-in-auto_smb.patch" not included?
[17:54] <rafaeldtinoco> bdmurray: we thought the & fix wasnt feasible as SRU
[17:54] <rafaeldtinoco> ahasenack and I is "we"
[17:54] <rafaeldtinoco> it isn't upstreamed yet (opened to discussion in mailing list)
[17:55] <rafaeldtinoco> and original bug was about admin shares (with $) only
[17:55] <rafaeldtinoco> so we thought about being conservative for & one
[17:56] <bdmurray> rafaeldtinoco: the groovy changelog reads like that patch is also for the same bug
[17:56] <rafaeldtinoco> its the same root cause perhaps
[17:56] <rafaeldtinoco> not the original reporter intent though
[17:57] <rafaeldtinoco> so I guess its okay as the discussion is all there
[17:57] <rafaeldtinoco> and both issues come from the same place (smb.auto)
[17:57] <rafaeldtinoco> auto.smb
[17:58] <bdmurray> rafaeldtinoco: what do you mean by "I guess its okay"? do you want to redo the upload?
[17:58] <rafaeldtinoco> bdmurray: https://bugs.launchpad.net/ubuntu/+source/autofs/+bug/1680224/comments/29
[17:59] <rafaeldtinoco> summarizes the situation
[17:59] <rafaeldtinoco> its okay to have LP for both patches
[17:59] <rafaeldtinoco> we opted to have the & fix only for groovy
[17:59] <rafaeldtinoco> and fix the $ as SRU
[18:00] <bdmurray> okay, if you decided to SRU the & fix then that'll require a new bug report
[18:00] <rafaeldtinoco> no problem! I'll put a comment in that bug mentioning that
[18:02] <rafaeldtinoco> bdmurray: added a comment and, if we ever decide to SRU & fix Ill open a new bug and refer to this one
[18:02] <rafaeldtinoco> thanks
[20:51] <mwhudson> kanashiro: argh i keep failing to get around to looking into that
[20:51] <mwhudson> kanashiro: yes, that debdiff is fine for now
[20:56] <kanashiro> mwhudson: thanks for taking a look at it. I will start to work on containerd 1.3.5 soon and try to follow what tianon suggested on the github issue regarding the library package
[20:57] <mwhudson> kanashiro: heh they released 1.3.6 already but yes
[20:58] <kanashiro> mwhudson: really?! So it will be 1.3.6 :)
[20:58] <mwhudson> kanashiro: i think it's some trivial ci fix
[21:02] <bdmurray> vorlon: How can I get my autopkgtest-cloud changes for scipy deployed?
[21:30] <bdmurray> jibel: Did you see that bug 1856574 was reopened?
[21:40] <vorlon> bdmurray: scipy> is there an MP link I missed?
[21:41] <bdmurray> vorlon: No, I just committed it. https://git.launchpad.net/autopkgtest-cloud/commit/?id=fc21a96fffc7cfac43c23150440d2a9b64c6518d
[21:41] <vorlon> bdmurray: ah
[21:42] <vorlon> bdmurray: it doesn't appear you are in the group for infra access despite having commit access, so I'll deploy it now
[21:42] <vorlon> bdmurray: (done, you can retry the tests now)
[21:43] <bdmurray> vorlon: thanks!