| hallyn | say, how come CONFIG_MEMCG_KMEM is diabled in trusty? | 01:56 |
|---|---|---|
| hallyn | (i'm sure there's a good reason, but i can't recall, or never knew) | 01:56 |
| === zz_nhayashi is now known as nhayashi | ||
| === sz0` is now known as sz0 | ||
| === sz0 is now known as sz0` | ||
| === Guest60783 is now known as hiddenbacon | ||
| === sz0` is now known as sz0 | ||
| === sz0 is now known as sz0` | ||
| === sz0` is now known as sz0 | ||
| === nhayashi is now known as zz_nhayashi | ||
| BjoernC | moin moin | 10:49 |
| ypwong | s | 10:49 |
| ypwong | oops | 10:49 |
| === psivaa is now known as psivaa-afk | ||
| === sz0 is now known as sz0` | ||
| === sz0` is now known as sz0 | ||
| hallyn | apw: hey, we were wondering how come CONFIG_MEMCG_KMEM is disabled in trusty? | 14:20 |
| rtg | hallyn, you might be wondering until Tuesday if you're gonna depend on apw to answer | 14:33 |
| rtg | hallyn, frankly, I'm not sure why CONFIG_MEMCG_KMEM is not enabled. Likely an oversight 'cause no body howled ? | 14:35 |
| hallyn | rtg: hm, ok, thx | 14:36 |
| hallyn | rtg: mainly it's the upstream-recommended way to prevent forkbombs, | 14:37 |
| hallyn | the reason they refuse to allow a 'ntasks' cgroup | 14:37 |
| rtg | hallyn, it kind of looks like something we should have enabled. how about starting a bug for it ? | 14:37 |
| hallyn | rtg: will do, thanks | 14:37 |
| wedgwood | I've been doing some testing to see about moving to Trusty from Debian and I'm seeing odd performance characteristics. | 15:10 |
| wedgwood | Seems that memory throughput is considerably slower on Ubuntu than anything else I've tried, but only for allocation sizes <512K | 15:10 |
| wedgwood | as measured with sysbench --test=memory | 15:11 |
| hallyn | rtg: filed bug 1309586 | 15:11 |
| ubot2 | Launchpad bug 1309586 in linux (Ubuntu) "enable CONFIG_MEMCG_KMEM" [Undecided,New] https://launchpad.net/bugs/1309586 | 15:11 |
| rtg | wedgwood, slub v.s. slab ? | 15:11 |
| rtg | hallyn, ack | 15:11 |
| wedgwood | rtg: I don't know what those are :( | 15:12 |
| rtg | memory allocators | 15:12 |
| wedgwood | is that dynamically tunable? | 15:12 |
| rtg | nope, they are configured at compile time. | 15:13 |
| wedgwood | rtg: whichever comes in the stock generic kernel. Is that something I can see with getconf? | 15:16 |
| rtg | wedgwood, egrep "SUB|SLAB" /boot/config-3.13.0-24-generic | 15:17 |
| wedgwood | # CONFIG_SLAB is not set | 15:18 |
| wedgwood | CONFIG_SLUB=y | 15:18 |
| wedgwood | The CentOS kernel I'm comparing with is SLAB. I'll check the debian one too | 15:19 |
| wedgwood | the debian kernel is SLAB, same as Ubuntu | 15:21 |
| wedgwood | so that doesn't seem to be the differentiator | 15:23 |
| wedgwood | *SLUB for debian, I meant | 15:31 |
| hallyn | rtg: thanks :) | 15:38 |
| caglar10ur | could any one why trusty comes with CONFIG_MEMCG_KMEM disabled? Was talking with hallyn yesterday about it and he suggested me to ask here | 15:55 |
| hallyn | caglar10ur: rtg has posted a patch to change that :) | 15:55 |
| caglar10ur | hallyn: ah nice :) | 15:56 |
| hallyn | see pad.lv/1309586 | 15:56 |
| hallyn | caglar10ur: i suspect it's something i should have noticed two months ago | 15:56 |
| caglar10ur | hallyn: great, I’m a happy person now | 15:56 |
| caglar10ur | hallyn: that’s also my fauly, I’ve been using trusty kernel for months and never mentioned that before :) | 15:57 |
| caglar10ur | hallyn: I was changing the config and repacking it | 15:57 |
| hallyn | heh | 15:58 |
Generated by irclog2html.py 2.7 by Marius Gedminas - find it at mg.pov.lt!