[01:37] PR snapd#9852 opened: gadget: enable multi-volume uc20 gadgets in LaidOutVolumeFromGadget; rename too [01:42] PR snapd#9847 closed: cgroup-support.c: fix link to CGROUP DELEGATION [07:53] is snapcraft release going to support releasing more that one revision at some point? [08:05] morning [08:05] pstolowski: hey [08:05] hey mborzecki [08:09] PR snapd#9840 closed: interfaces/builtin: Allow DBus property access on org.freedesktop.Notifications [08:59] PR snapd#9849 closed: tests/main/snap-network-errors: skip flushing dns cache on centos-7 [09:59] PR snapcraft#3425 opened: schema: add support for activates-on app property to schema === msalvatore_ is now known as msalvatore [12:49] hey mborzecki [12:49] ijohnson: hey [12:49] do you have thoughts on whether the sector size for i.e. uc20 should be dynamically determined at runtime or if we should just have the gadget specify it per-volume ? [12:50] having it specified in the gadget is potentially easier but also potentially requires more design input and so could take longer than just figuring it out ourselves when we need it [12:50] ijohnson: the problem you observed yesterday happened when installing uc20 on internal emmc right? [12:50] mborzecki: yes [12:52] ijohnson: hm it's not only that, if we increase block size, the size of journal will increase as well, so there will be less free space in ubuntu-save [12:52] s/free/usable free/ [12:52] yeah I was wondering about that as well, but my initial patch just moves the block size up to the sector size if the sector size is specified [12:52] so for example on amd64 devices you wouldn't see any change [12:53] but on a device that has a sector size that is larger than 1024 (i.e. 4096 like this device) then we would set the block size [12:54] ijohnson: yeah, i suppose we could it this way [12:54] ijohnson: maybe we should have a check for some about of free space in ubuntu-save too [12:54] ijohnson: since the adjustment would happen at runtime, we should check it at that time too [12:54] yeah we could add some validation that is like "if sector size > 1024 && size < 10M" or something like that [12:55] mborzecki: what were your thoughts on dynamically detecting the sector size? I seem to recall something with using ioctl or something like that and it being racy? [12:55] ijohnson: iirc it's worse than that, with 4k block size journal is at least 4M, if you add encryption on top, maybe you'll have 5M of free space on 10MB partition [12:56] mborzecki: yeah I just meant whatever numbers make sense, I don't know what numbers make sense for such a check :-) [12:56] ijohnson: there was an ioctl you could use, but I don't think it's exposed in to go [12:57] just a sec, it was in the e2fsprogs code i linked yesterday [12:58] ijohnson: BLKSSZGET [12:59] cool [13:01] now we only need to device on the amount of free space we expect in ubuntu-save [13:03] mborzecki: seems that `sudo blockdev --getbsz /dev/sda` works and blockdev is in the core20 snap [13:04] ijohnson: ah, that's nice, can you also check what sfdisk reports for that block device? [13:04] sfdisk doesn't show the block size do you just want the sfdisk json ? [13:04] err s/block size/sector size/ [13:04] https://www.irccloud.com/pastebin/EzSaIHf3/ [13:04] ijohnson: both --dump and --json [13:04] ^ [13:05] ijohnson: and --dump? :P [13:05] seems the same? [13:05] https://www.irccloud.com/pastebin/zNXzZeLf/ [13:06] I ran `sudo sfdisk --dump --json /dev/sda` [13:06] ijohnson: just `sfdisk --dump` [13:06] ah I see what you mean [13:07] ijohnson: and if there's parted, please run `parted p` too [13:07] https://www.irccloud.com/pastebin/GgwC8Q30/ [13:07] no parted, let me see if I can quick whip that into a devmode snap [13:08] ijohnson: funny how sfdisk doesn't even include sector size, maybe it's an older version, cause i see it in the output locally [13:11] yeah seems odd to me, the version of sfdisk I have in groovy outputs the sector-size as well [13:11] wait wat [13:11] ``` [13:11] sfdisk: mutually exclusive arguments: --list-free --json --dump [13:11] ``` [13:19] mborzecki: I got parted to run if you are curious [13:19] https://www.irccloud.com/pastebin/QSXfTkc4/ [13:19] ha, nice [13:19] ijohnson: can you run `parted unit s p` ? [13:20] https://www.irccloud.com/pastebin/GKAlsL6o/ [13:21] ijohnson: hah [13:21] ijohnson: so, sfdisk reports the sizes in units of 4096 [13:21] ijohnson: and offsets too for that matter [13:22] right [13:25] ijohnson: are the partition sizes correct, do they match the gadget? [13:25] let me double check but yes they should be, since I just s/512/4096/ for all the sector sizes in snapd to get this board to install :-) [13:27] mborzecki: yeah it all lines up [13:29] hahah [13:29] wow, it's a bummer though that sfdisk does not report sector size it's using [13:30] yeah really sad, but I think using the blockdev command I showed above is probably good enough [13:30] and it's apparently the logical sector size that matters, parted and sfdisk show correct info on my 512/4096 disk [14:31] winter is so much fun, -25C on tuesday, +5C right now [14:35] indeed [15:01] * cachio lunch [15:40] good afternoon [15:41] * zyga was AFK for good part of the day [15:41] hey zyga [15:41] ijohnson, hey, I'll tar that up today, sorry, I had to be in the office all morning [15:41] no worries === tomreyn_ is now known as tomreyn [15:55] PR snapcraft#3426 opened: Ensure PYTHONPATH is properly set for gnome-3-34 builds === strohi is now known as strohalm === ijohnson is now known as ijohnson|lunch === ijohnson|lunch is now known as ijohnson