nyov | cking: hi! once upon a time, you wrote this answer: https://askubuntu.com/a/729758 - Are you (or anyone else) aware if this is fixable via ACPI (DSTD) table patches? | 10:35 |
---|---|---|
sdhd-sascha | Hi, does anybody know, how to ensure that a zfs directory is synced to disk. I have trouble with `du` (directory usage size) | 15:27 |
tomreyn | i assume "sync" doesn't then? | 16:13 |
cking | nyov, let me have a look | 16:22 |
nyov | cking: oh awesome. The relevant code would be here https://github.com/torvalds/linux/blob/master/arch/x86/kernel/apic/io_apic.c#L2210 | 16:24 |
cking | sdhd-sascha, the du command won't give you the correct info, you need to use zfs list | 16:25 |
cking | zfs knows more about how much is really being used (because of snapshots etc) than du can provide | 16:26 |
cking | nyov, i doubt that dabbling with tweaking the DSDT will be an easy route to fixing this, best try the kernel parameter 'noapic' as as workaround | 16:27 |
sdhd-sascha | cking: thank you. | 16:35 |
sdhd-sascha | The problem is, that "ubuntu-image" creates a temporary directory and `du` calculates the rootfs size. On my system i use /tmp/ on zfs with "sync=always" enabled | 16:35 |
sdhd-sascha | Is it possible to use "zfs list" to get a subdirectory size ? I also tried `sync -f $(find /tmp/subdir)` . But didn't help in the most cases | 16:37 |
sdhd-sascha | I also have create extra ZIL and L2ARC on a SSD ... | 16:39 |
sdhd-sascha | cking: or is there a system-call which i could call, to get the correct size ? | 16:42 |
nyov | cking: thankfully thats not necessary, the kernel finds the right workaround and it works. I was just curious if I could fix it, since I already have to fix so much else in the DSDT. | 16:43 |
nyov | (stupid notebook can't even wake up from standby) | 16:44 |
sdhd-sascha | @tomreyn: right, sync did not work immediately | 17:03 |
sdhd-sascha | After some seconds the dir size is corrected | 17:04 |
sdhd-sascha | ... just read some solaris posts. I think i can imagine the problem with COW and compression now | 17:09 |
sdhd-sascha | Shortest solution could be to iter all files and sum up the size. Or? | 17:10 |
sdhd-sascha | Would linux sys_newstat return correct values in all cases? E.g. compression | 17:39 |
sarnold | sdhd-sascha: have you seen this yet? https://lists.freebsd.org/pipermail/freebsd-stable/2010-May/056654.html | 18:47 |
sdhd-sascha | sarnold: not yet. What is the best place for kernel interface documentation ? i found something like "kernel.readthedocs.org" | 18:51 |
sdhd-sascha | thank you :-) | 18:51 |
sdhd-sascha | My last reading about kernel programming was around 2.2 ~ 2.4 ... | 18:51 |
sarnold | sdhd-sascha: there's no single good answer for kernel interface docs; the source code is the source of truth but also hard to digest; the zfsonlinux documentation is among the best documentation around, but is of course a bit specific to its own needs | 18:52 |
sdhd-sascha | okay :-) grep'ing the headers :-) | 18:53 |
sdhd-sascha | sarnold: just read the thread from the mailing-list. great, thank you, again +1 | 19:24 |
sarnold | sdhd-sascha: woo, cool :) | 19:25 |
tqinli | Hi, I file this bug report 6 months ago: https://bugs.launchpad.net/ubuntu/+source/linux/+bug/1835940, wondering what is the plan to have the fix backport to 4.15 kernel ? | 20:41 |
ubot5 | Ubuntu bug 1835940 in linux (Ubuntu) "perf core dump at tool/perf/util/namespaces.c:144" [Undecided,Confirmed] | 20:41 |
Generated by irclog2html.py 2.7 by Marius Gedminas - find it at mg.pov.lt!