=== gerald is now known as Guest88751 | ||
pkern | Hi. What's the timeline for 3.16.0-38.52~14.04.1 to be pushed to trusty proper? | 08:40 |
---|---|---|
henrix | pkern: that kernel is currently under testing/verification. if everything goes according to the plan, it should be out of -proposed by the end of this week | 08:45 |
henrix | (we're currently on the 3rd week of the 3 weeks SRU cycle) | 08:46 |
infinity | It actually just needs a security sign-off, and it could go out today. | 08:55 |
infinity | jjohansen: Any chance you'll get around to kernel security signoffs today? | 08:56 |
dileks_webchat | hi | 11:18 |
dileks_webchat | I have a hdd in a notebook and want to deactivate hdd-scan when using a recent ubuntu live-system. do you happen to know if there is a boot-option for that? | 11:19 |
dileks_webchat | I checked <http://wiki.ubuntuusers.de/Bootoptionen?redirect=no> (German) | 11:20 |
dileks_webchat | cannot see a suitable boot-option | 11:20 |
apw | dileks_webchat, i don't think i know what you are referring to when you say hdd-scan | 11:23 |
apw | exactly what are you trying to prevent occuring ? | 11:23 |
dileks_webchat | the hdd is defect and should not be considered for installation or whatever. handled as non-existent | 11:24 |
dileks_webchat | hdd is full of s.m.ar.t errors | 11:26 |
dileks_webchat | hardware-detection on boot-up results in a restart of the live-system. hdd cant be deactivated in bios. | 11:29 |
apw | dileks_webchat, ok makes sense, i am supprised it would trigger a restart, i would expect it to just take longer to boot | 11:32 |
apw | dileks_webchat, as i have used live images often to get what is still readable off broken laptop drives in the past | 11:32 |
apw | dileks_webchat, it is possible something like libata.force=disable might do what you want | 11:34 |
apw | you might need to limit it to the device whihc is in error if that also gets rid of your usb stick | 11:35 |
dileks_webchat | how do I limit to the device - which parameter do I pass? | 11:37 |
apw | dileks_webchat, never done it, it just says [ID:] in front of the disable; | 11:38 |
apw | libata.force= [LIBATA] Force configurations. The format is comma | 11:38 |
apw | separated list of "[ID:]VAL" where ID is | 11:38 |
apw | PORT[.DEVICE]. PORT and DEVICE are decimal numbers | 11:38 |
apw | matching port, link or device. Basically, it matches | 11:38 |
apw | the ATA ID string printed on console by libata. If | 11:38 |
dileks_webchat | hehe, I was looking at that :-) | 11:39 |
dileks_webchat | http://git.kernel.org/cgit/linux/kernel/git/torvalds/linux.git/tree/Documentation/kernel-parameters.txt#n1745 | 11:39 |
dileks_webchat | apw: thanks for the quick answer! | 11:42 |
dileks_webchat | bye | 11:47 |
mickkelodeon | hi | 11:54 |
mickkelodeon | I don't know if this is the best place for asking about low-latency/real-time kernels with ubuntu and ec2 instances... | 11:55 |
apw | mickkelodeon, without knowing the question that is hard to say | 12:00 |
mickkelodeon | I have more than 100 processes working in real time, and the latency is killing me | 12:01 |
mickkelodeon | I want to change the kernel for another with better performance for this purpose | 12:01 |
mickkelodeon | I was taking a look to the lowlatency kernel, but when I installed it I got "Ignoring non-Xen Kernel on Xen domU host: vmlinuz-3.2.0-82-lowlatency" | 12:11 |
smb | iirc that is keyed on some CONFIG_XEN ... could be the default options for lowlatency differ there | 12:13 |
mickkelodeon | because the realtime kernels seem outdated | 12:14 |
apw | smb, i would be supprised if we changed xen in lowlatency | 12:22 |
apw | smb, unless it is somehow contra-indicated of course for the other options | 12:23 |
smb | The realtime patchset was a big effort to maintain and nobody cared to fund it. I believe I remember vaguely that there was some recent decision to bring it upstream | 12:23 |
smb | apw, Yeah, I am not sure where lowlatency came back then | 12:23 |
apw | smb, yeah thats my memory that RT is going to die either by merging or dissappearnig | 12:23 |
smb | then it also could be the way the special update-gub checks. That also has changed over time from checking kernel names to kernel options... | 12:25 |
smb | If I could remember all those dates... :/ | 12:25 |
smb | Hrm, the responsible package also changed name and place. Used to be grub-lecacy-ec2 (part of cloud-init source) and now might be pv-grub-menu since Trusty | 12:29 |
mickkelodeon | so, smb apw what is the best way in my case? | 12:43 |
pkern | henrix: Thanks! | 12:45 |
smb | mickkelodeon, I am looking at update-grub-legacy-ec2 which comes via grub-legacy-ec2 in Precise. If I cannot find a nicer way, maybe fiddling with is_xen_kernel could work for you... though probably a bit of a pita if it needs doing for multiple installs/instances | 12:46 |
smb | Its bascially a bash script and as I suspected it keys on the -virtual part of the kernel filename back then | 12:47 |
apw | smb, ug | 12:49 |
mickkelodeon | smb , but do you think this could be a substantial change in terms of performance for my case? | 12:49 |
smb | Hm, maybe one could even define the shell function seperately as it checks for it to exist | 12:49 |
apw | audio peoiple claim it helps latency significantly for their use case, but ymmv | 12:50 |
smb | Yeah... there is a catch maybe for in P -virtual may have had more essential things build in for running under Xen than generic | 12:51 |
apw | smb, ugg | 12:52 |
smb | apw, Yeah... with lucid gone now we went from quite different to subtly different which is hard to remember | 12:53 |
apw | smb, yeah | 12:53 |
trippeh | I doubt a realtime kernel would make much difference on AWS/ec2 | 12:53 |
trippeh | on public clouds you often have to build your system to be tolerant to some extra latency | 12:55 |
smb | with the addition of vcpu scheduling its hard to predict ... i guess | 12:55 |
smb | So it might be that lowlatency improves a little by being preemptive and setting a higher HZ because it lowers a single tasks effect but still a vcpu can stall or cause latency outside the kernels control | 13:06 |
jjohansen | infinity: yes | 15:02 |
=== kamal__ is now known as kamal | ||
cyking | i've been asked to fully commit bisect the kernel. can someone help walk me through the process. thanks. | 19:45 |
cyking | https://bugs.launchpad.net/ubuntu/+source/linux/+bug/1453391 | 19:45 |
ubot5 | Launchpad bug 1453391 in linux (Ubuntu) "Very slow Qualcomm Atheros AR8121/AR8113/AR8114 Gigabit or Fast Ethernet (rev b0)" [Medium,Incomplete] | 19:45 |
cyking | seems that I should focus on one of these comnits http://pastebin.com/WwVjC02W but the instructions are not so clear to me. | 20:36 |
cyking | trying to build kernel but debian.master and debian folders do not exist. what to do ? | 21:45 |
=== mukhbiir is now known as neoark |
Generated by irclog2html.py 2.7 by Marius Gedminas - find it at mg.pov.lt!