[00:59] PR snapd#6625 opened: tests: system disable ssh for config defaults in gadget [01:01] is there a way via the snapd api to see if refreshing to a particular risk (in particular: including a branch) will perform an upgrade? [01:07] mwhudson: "risk (in particular: including a branch)" is just a channel [01:07] wgrant: well ok my question still stands [01:07] mwhudson: snap refresh --list SNAP --channel=CHANNEL should do it, but it's not supported [01:08] Not sure why [01:08] --list doesn't take the normal refresh args [01:09] it's a very different path inside snapd for whatever reason [01:09] you can say snap switch --channel=foo SNAP, snap refresh --list [01:09] Ah, --list uses find [01:09] That's weird [01:09] but that seems to silently succeed if you use a bogus channel [01:10] Yeah, snap refresh --channel=foo SNAP checks that the target channel exists, but snap switch doesn't [01:11] It doesn't look like snapd provides a sufficient API :( [01:12] heh switch to a missing channel and refresh --list does end up with a complaint in the logs [01:12] but i don't think i should be that evil :) [01:13] snap refresh --no-wait --channel=foo SNAP and assume i can abort the job in time? :) [01:15] Heh [05:58] morning === chihchun_afk is now known as chihchun [06:23] shall we land #6620? [06:23] PR #6620: tests/main/remodel: clean up before reverting the state [06:51] Good morning [06:51] Hey Maciej [06:52] mborzecki: I will likely have a slower day today [06:52] Looking at 6620 [06:52] zyga: hey hey [06:53] Yeah [06:53] Let’s land it [07:32] good morning mvo [07:32] I just merged the fix from chipaca [07:32] let's see some green today :-) [07:33] PR snapd#6620 closed: tests/main/remodel: clean up before reverting the state [07:33] hey zyga [07:33] I, selfishly restarted https://github.com/snapcore/snapd/pull/6622 - let's see if it goes green now [07:33] hey :) [07:33] zyga: thank you [07:33] PR #6622: cmd/libsnap: rename C enum for feature flag [07:36] zyga: lets focus on the 2.38 tagged one(s) initially :) [07:37] I'll wait and see if anything turns green [07:37] mvo: https://github.com/snapcore/snapd/pull/6617 needs a review from You [07:37] PR #6617: cmd/snap: fix regression of snap saved command <⚠ Critical> [07:47] PR snapd#6607 closed: cmd: typedef mountinfo structures === pstolowski|afk is now known as pstolowski [08:02] mornings [08:06] hey pstolowski [08:16] hey pawel :) === chihchun is now known as chihchun_afk === chihchun_afk is now known as chihchun [08:35] hmm [08:35] tests still red? [08:36] PR snapd#6626 opened: strutil: make SplitUnit public, allow negative numbers [08:36] should be a simple one ^^ [08:36] pstolowski: mvo: hey [08:37] mborzecki: good morning [08:38] mvo: https://github.com/snapcore/snapd/pull/6613 needs a 2nd review [08:38] PR #6613: interfaces/builtin: add dev/pts/ptmx access to docker_support (for 2.38) [08:38] zyga: on it [08:39] mborzecki: is there anything that unicode.IsDigit returns true to that is not in range '0'..'9'? [08:39] like any funky numerals? [08:40] zyga: iirc that's what IsNumber could do [08:40] afaik IsDigit is 0-9 [08:40] mborzecki: https://golang.org/src/unicode/digit.go [08:41] mborzecki: https://golang.org/src/unicode/letter.go line 170 [08:41] mborzecki: I think it does more than you think it does :) [08:42] at least my naive understanding would suggest that there are more characters that match [08:42] can you run a simple loop over 65K runes and see what matches? [08:43] mborzecki: https://golang.org/src/unicode/tables.go line 2308 [08:43] there's a lot of digits [08:43] pfff [08:44] ৬ hello there little digit [08:45] whaat? what is that? [08:45] google says that is 6 [08:45] well [08:45] I think the branch needs some more changes [08:45] 'bengali digit one' [08:46] 一 [08:46] I wonder if that also matches [08:46] zyga: but it doesn't parse as a number [08:46] 二 [08:47] "߁kB" -> cannot parse \"߁kB\": \"\\xdf\" is not a number [08:47] hmm [08:47] that's not what I sent, perhaps IRC client mangles that [08:48] anyway :) [08:50] PR snapd#6613 closed: interfaces/builtin: add dev/pts/ptmx access to docker_support (for 2.38) [08:52] tests are still ungreen [08:52] fatal: unable to access 'https://github.com/snapcore/squashfuse/': gnutls_handshake() failed: Error in the pull function. [08:52] eh [09:02] hey chihchun [09:02] hey Chipaca [09:03] hey tab completion :/ [09:03] hey zygurat [09:03] :-p [09:03] zyga: 'sup? [09:03] Chipaca: it's a sunny day [09:03] that's always good [09:03] * Chipaca looks outside [09:03] lies [09:07] zyga: mborzecki: i agree about the unicode.IsDigit, saw that the other day and added a todo to fix it [09:07] zyga: mborzecki: but in my case i was going to change it to return uints everywhere [09:08] (because we weren't using it from anywhere that needed negative sizes anyway) [09:08] mborzecki: but [09:08] mborzecki: we rely on the behaviour of never returning negative [09:09] ah but it now checks for that explicitly [09:09] ok :-) [09:13] hm have to fix the slicing too, otherwise it'll split up the unicode chars incorrecty [09:14] mborzecki: ? [09:14] Chipaca: just antoher bug in the PR that i see now [09:15] mborzecki: did you change the loop from a range over the string? [09:15] mborzecki: ranging over a string yields full runes [09:16] PR snapd#6627 opened: devicestate: deal correctly with the "required" flag on Remodel [09:20] mborzecki: https://play.golang.org/p/SUc1sdPpeUd fwiw [09:22] mborzecki: or https://play.jsgo.io/8e1f7704efa0f1ca794f16e88134281cf870fe93 :-p [09:23] * Chipaca stops having fun and gets to work [09:23] PR snapd#6628 opened: overlord/snapshotstate: support auto flag (automatic snapshots 1/N) [09:24] jsgo.io [09:24] that's quite fancy [09:24] and 3rd party imports work too [09:25] mborzecki: they even tweaked the limits so we could import snapd in there [09:25] (i mentioned this when it happened) [09:25] Chipaca: zyga: updated the PR, please take a look [09:25] enqueued, doing another review now [09:26] zyga: #6605 needs master merge [09:26] PR #6605: cmd/libsnap,osutil: fix parsing of mountinfo [09:27] mvo: hi, could you re-review #6575 when you have some time [09:27] PR #6575: cmd/snap-confine: pass sc_invocation instead of numerous args around [09:27] mborzecki: thank you, I will de-conflict that soon [09:27] need to grab some coffee [09:29] relative offsets in gadget.yaml are so fancy [09:31] mborzecki: I'd expect you to make a ParseByteSize analogue, ParseNonISOByteSize (or something), instead of exporting the helper, fwiw [09:32] mborzecki: but I don't mind (obviously) [09:32] * Chipaca not that worried [09:35] * zyga could use a coffee too but first a quick merge with master [09:35] Chipaca: would like that, but gadget takes sizes in format 1/1M/1G, SplitUnit is all i need [09:42] https://github.com/snapcore/snapd/pull/6614 needs a 2nd review now [09:42] PR #6614: cmd/snap-confine: use fixed private tmp directory [09:42] mborzecki: conflict resolved, thanks [09:43] wodner how did google come up with 'stadia' as the name [09:45] mborzecki: https://www.dailymotion.com/video/x1afila [09:47] mborzecki: (seek to the end unless you know this one ;) [09:49] heh [09:53] * zyga looks at https://forum.snapcraft.io/t/problems-installing-a-base-core18-snap-on-bionic/10084/2 [09:53] but first coffee, it's so cold in the office still [09:53] I need to print something again, it's always cozy when that thing is on [09:59] Chipaca: can we have a chat about 6624, everything is more complicated than we remembered [10:01] pstolowski: #6628 includes #6617 ? [10:01] PR #6628: overlord/snapshotstate: support auto flag (automatic snapshots 1/N) [10:01] PR #6617: cmd/snap: fix regression of snap saved command <⚠ Critical> [10:01] pedronis: sure [10:01] Chipaca: no, i'm waiting for 6617 to become green, will merge and deconflict [10:01] pedronis: but if it really is, that means it isn't 2.38 material, so I vote we refactor now and fix it properly [10:02] Chipaca: well the issue is not so much the change to autorefresh, is the store changes [10:03] and maybe i'm very optimistic hoping for green.. [10:03] pedronis: i'm listening [10:03] pstolowski: what's red now? [10:03] Chipaca: can we do a quick HO? [10:03] pedronis: sure, in 5? [10:03] Chipaca: works for me, thanks [10:05] Chipaca: just saying, still a couple of red PRs from there, maybe just need restarting [10:05] d/from/ [10:06] * Chipaca searches for his headset [10:09] pedronis: ready when you are [10:10] Chipaca: omw [10:16] pedronis: sure, will re-review. thank you [10:44] oh noes [10:45] Chipaca: you asked about what's red... #6617 went red [10:45] PR #6617: cmd/snap: fix regression of snap saved command <⚠ Critical> [10:45] fatal: unable to access 'https://go.googlesource.com/net/': Failed to connect to go.googlesource.com port 443: Connection timed out [10:46] pstolowski: ^ [10:46] lots of timeouts from there [10:46] pstolowski: restarted [10:46] pstolowski: thank you [10:57] pedronis: maybe i should've split the the backoff test, but it exercises the bits we want it to now [10:57] Chipaca: if you think two tests make more sense, yes let's do that [10:58] pedronis: i think one is fine, even though the name could use a tweak [10:58] as it's not just about the backoff [10:58] but the comments cover what's going on [10:58] and they're all about backoff in one way or another, so maybe it's fine [10:58] Chipaca: maybe it should have Error in the name? [10:59] is is about errors after all [11:00] Chipaca: ty! [11:15] zyga: I need to have lunch, we should probably have a chat about 6597, seems we are a bit talking across each other [11:24] PR snapd#6622 closed: cmd/libsnap: rename C enum for feature flag [11:25] pedronis: gladly, I would prefer to do it before the standup if you can because I need to break early today [11:44] zyga: 20 mins before the standup? === chihchun is now known as chihchun_afk [11:47] zyga, pedronis quick question abut https://github.com/snapcore/snapd/pull/6575#discussion_r264631829 then I can continue with the review [11:47] PR #6575: cmd/snap-confine: pass sc_invocation instead of numerous args around [11:47] zyga: but I'm puzzled by this one [11:47] * mvo probably miss something obvious [11:51] mvo: looking [11:52] mvo: replied [11:52] pedronis: sounds good [11:54] mvo: the move make the unit test linking fail, I suppose it's .h bit of the move [11:54] s/linking/compiling/ [11:56] * mvo nods [11:59] pedronis: we could move it again but I would have to rework snap-confine.c to effectively move main somewhere else [12:10] PR snapd#6629 opened: overlord/snapshotstate: helpers for dealing with expiration times [12:10] pedronis, hey, I see this error on the new test to disable ssh [12:10] https://paste.ubuntu.com/p/d9ckjJWMPk/ [12:11] cachio: we don't mask ssh, we use a file instead [12:11] mvo: ^ [12:12] pedronis, mvo which file? [12:13] one sec [12:15] cachio: /etc/ssh/sshd_not_to_be_run [12:15] pedronis, nice, thanks [12:15] cachio: code is here: https://github.com/snapcore/snapd/blob/master/overlord/configstate/configcore/services.go#L59 [12:16] we do unmask in case when (re)enabling, but our own disabling logic is based on that file [12:16] there's a conditional in the unit [12:18] pedronis, reading, thanks for the info [12:20] pedronis, cachio its complicated, we used to "mask" but had bugs becaus of ssh/sshd aliases and the recommended way was to use this file [12:21] pedronis: re your comment to measurements PR about capturing task status, that's a good idea; does "task-status" tag sound ok? [12:22] zyga: reviewed, sorry for the delay, had some questions [12:22] pstolowski: yes [12:22] mvo, it is ok, I with that code I can adjust the test case [12:22] mvo: no worries [12:22] pstolowski: like task-kind [12:22] cachio: thanks [12:27] pedronis: ty [12:34] and now google:fedora-29-64:tests/main/degraded failed. fun. [12:37] * pstolowski lunch [12:44] mvo: updated https://github.com/snapcore/snapd/pull/6575 - I left one thing out (cleanups memset 0) and I would prefer to do it in a separate branch [12:44] PR #6575: cmd/snap-confine: pass sc_invocation instead of numerous args around [12:48] mvo what's your plans for lp #1819728? i have a couple/few systemd patches i'm planning to sru soon, i can include your patches if you think they will be ready this or next week [12:48] Bug #1819728: Please backport "fix race between daemon-reload and other commands #8803" to 16.04 (for UC16) and 18.04 (for UC18)

Bionic):Fix Committed> [12:49] also i assume the current systemd bionic upload will be rejected due to failed verification, so you'll need to include lp #1778936 patch in the next upload as well [12:49] Bug #1778936: please re-add Support-system-image-read-only-etc.patch

[13:12] off to pick up the kids, may be a bit late to the standup [13:21] mborzecki: https://github.com/snapcore/snapd/pull/6626 is green [13:21] PR #6626: strutil: make SplitUnit public, allow negative numbers === ricab is now known as ricab|lunch [13:48] ddstreet: that sounds reasonable, I should have things ready by then. its not at all straightforward unfortunately because its 239->229 and there are a bunch of changes which cause conflicts etc. plus its a delicate area of the code. but I think i have something ready [13:48] ddstreet: do you have a link to your things? just to double check that we don't conflict? [13:51] mvo so far we have lp #1812760 and lp #1818282 which we're getting patches ready for, but also 2 more issues without bugs yet that we're investigating [13:51] Bug #1812760: networkd: [Route] PreferredSource not working in *.network files Progress by ddstreet> [13:51] Bug #1818282: systemd-networkd - RoutingPolicyRule does not apply correctly

Progress by joalif> [13:52] mvo also, there may be a systemd security release coming soon so we might be waiting until after that to upload [13:53] ddstreet: thanks, that looks like a totally different area, so that should be fine [14:13] PR snapd#6617 closed: cmd/snap: fix regression of snap saved command <⚠ Critical> [14:34] cachio: spread -list google:[^u]...:tests/... vs spread -list google:[u]...:tests/... [14:51] pedronis: updated https://github.com/snapcore/snapd/pull/6597 [14:51] PR #6597: cmd/snap-update-ns: refactor of profile application (1/N) [14:54] Bug #1821023 opened: core18 base on core 16 missing firmware [15:06] Is there an API call to determine which base a snap has before installation? [15:08] tobikoch: yes, see what happens when you `snap info --verbose jq-core18` === ricab|lunch is now known as ricab [15:22] zyga: I see, I have to explicitly request the field. [15:22] Chipaca, nice, thanks [15:22] yes, I believe so [15:22] * zyga needs to break, need to study before classes === mborzeck1 is now known as mborzecki [15:31] tobikoch: request it where? [15:33] Chipaca: refresh API endpoint [15:33] tobikoch: on api.snapcraft.io? [15:33] Chipaca: yes [15:34] tobikoch: refresh endpoint, or info endpoint? [15:34] Chipaca: refresh [15:34] strange, but ok [15:34] tobikoch: what're you doing? [15:35] (out of pure unadulterated curiosity) [15:35] Updating the snap download tool in livecd-rootfs [15:36] tobikoch: 'snap download' not cutting it? [15:37] Chipaca: snap download was fine, but it doesn't allow passing along a cohort key. [15:37] So until it does, there is this script. [15:37] * cachio lunch [15:38] pedronis: ^ were you aware that livecd-rootfs was using cohorts already? [15:38] tobikoch: fascinating :-) [15:38] heh [15:38] tobikoch: if you're doing bases like that, how're you doing other prerequisites (content providers, particularly)? [15:38] Chipaca: yes [15:39] I was aware [15:39] Chipaca: not sure I understand, so far the script does what it needs, I may be missing out on the wider scope. [15:39] tobikoch: if it's not a proble, it's not a problem :-) [15:39] :) [15:40] Chipaca: this is only for download, it does nothing else. [15:40] It's used for snap-preseeding in our images. [15:40] Chipaca: they might need changes, but that's really a discussion to have with the store for them [15:41] k [15:43] pedronis: actually I had mentioned it here: https://forum.snapcraft.io/t/managing-cohorts/8995/2 [16:00] zyga: could you re-review #6485 (it has two reviews but it had a complicated history so a 3rd will not hurt) [16:00] PR #6485: interfaces/seccomp: regenerate changed profiles only [16:00] pedronis: certainly! [16:04] thx [16:16] Hi. maybe someone can help me: my apache user www-data should run a snap app (like /snap/bin/vlc) but gives me „cannot create user data directory: /var/www/snap/vlc/770: Read-only file system“ - first: the directoy is created by this app (its there), but second: could it be that there are missing permissions (solved with snap connect) ? [16:17] iTommix: hey, just a partial answer, read only file system means that you are on a read only file system, permissions cannot fix this [16:17] iTommix: I will let others help you, I'm partially AFK / commutting [16:18] zyga: thats not right… the FS is writable. [16:18] is the path realy /var/www/snap/vlc/770? [16:18] that's unusual [16:19] yes, because of the user www-data home-dir its /var/www [16:19] i have an own „snap“-folder in my homedir [16:20] iTommix: please seek help on the forum, I'm sorry I cannot help you more now [16:20] will wait till others answer maybe… thanks anyway :) [16:24] PR snapd#6610 closed: interfaces/builtin: add add exec "/" to docker-support [16:25] PR snapd#6630 opened: cmd/snap: make 'snap warnings' output yamlish [16:47] pedronis: do you want to squash merge 6624 and tweak the commit message or shall I do it? the current commit message does not match the reality of the PR anymore (or maybe Chipaca can do it when he is back?) [16:47] mvo: I'm about to go afk [16:47] ah [16:47] mvo: i'll do it now [16:47] so I think either you or John [16:48] Chipaca: thx [16:48] Chipaca: thank you! later is fine if you are on the run [16:48] mvo: done [16:49] about to disappear but here still :-) [16:51] Chipaca: \o/ [16:52] PR snapd#6624 closed: overlord/snapstate: retry less for auto-stuff [17:20] PR snapd#6631 opened: tests: split travis spread execution in 2 jobs for ubuntu and non ubuntu systems [17:23] pedronis, Chipaca just FYI, I added a comment in https://bugs.launchpad.net/snapd/+bug/1768419/comments/4 [17:23]