| === rsalveti_ is now known as rsalveti | ||
| === JaMa|Wrk is now known as JaMa | ||
| === rsalveti__ is now known as rsalveti | ||
| _Kamondelious_ | **tumbleweed** | 18:49 |
|---|---|---|
| cwillu | _Kamondelious_, it's saturday, what did you expect? | 22:03 |
| cwillu | everybody is actually playing with their arm-based devices today | 22:03 |
| rcn-ee_lpt | cwillu, saw your irc message, did you have any luck finding that oom defrag config option? | 22:04 |
| cwillu | sec | 22:05 |
| cwillu | not specifically for oom | 22:05 |
| rcn-ee_lpt | no rush. ;) | 22:05 |
| cwillu | it's a case of "Carey saw a random thing added in the latest kernel, and must have it" :p | 22:05 |
| cwillu | memory compaction | 22:07 |
| rcn-ee_lpt | well... if linus does do a rc2 today, it could be avaiable in the next rc2 build.. ;) | 22:07 |
| cwillu | CONFIG_COMPACTION | 22:08 |
| cwillu | relies on CONFIG_MIGRATION I believe, although I guess that'd get set automatically | 22:09 |
| cwillu | also, did you see my note about 2.6.34-l0? | 22:09 |
| cwillu | I respun my image against 2.6.34-l1 before I left work yesterday, logging in right now to see if it's suffering the same fate | 22:10 |
| cwillu | and it appears that it might | 22:10 |
| rcn-ee_lpt | the oom-killer thing? http://paste.pocoo.org/show/221947/ | 22:11 |
| cwillu | yep | 22:11 |
| cwillu | I'm seeing a very slow response over ssh, about a minute to get a login prompt, and 15 seconds or so to verify a password | 22:12 |
| cwillu | I've got a serial terminal hooked to it, which seems a bit more responsive | 22:12 |
| rcn-ee_lpt | very strange, i haven't hit that yet... how much are you loading your systems? | 22:12 |
| cwillu | sec, pulling up top over the serial link | 22:13 |
| cwillu | it's just firefox with a backend server, streaming data to the page | 22:13 |
| cwillu | what I see is basically a swapstorm (noting that I don't have any swap configured on the system) | 22:13 |
| rcn-ee_lpt | so then it shouldn't be too much.. | 22:13 |
| cwillu | there shouldn't be any io load to speak of | 22:14 |
| cwillu | okay, it's definitely an issue under l1 as well | 22:14 |
| rcn-ee_lpt | okay, that's the one difference i do have swap setup on my test ones.. | 22:14 |
| cwillu | top just came up finally, I'm seeing the same sort of symptoms | 22:14 |
| cwillu | also, it doesn't show up for the first hour or so | 22:14 |
| cwillu | I've got high wait% showing in top, with btrfs-endio-0 and kswapd0 showing | 22:15 |
| rcn-ee_lpt | and looking at irc, it shows there is memory left, but it's going nutz.. | 22:15 |
| cwillu | sorry, looking at irc? | 22:16 |
| cwillu | looking at yours you mean? | 22:16 |
| rcn-ee_lpt | sorry irc log from yesterday.. ;) i'll setup a quick btrfs and no swap rig, and let it run for an hour to see what happens.. | 22:16 |
| cwillu | 6.85 load average :) | 22:17 |
| cwillu | oh, here's something | 22:17 |
| cwillu | Cpu(s): 10.1%us, 22.3%sy, 0.0%ni, 0.0%id, 67.3%wa, 0.0%hi, 0.3%si, 0.0%st | 22:17 |
| rcn-ee_lpt | that's way more.. | 22:17 |
| cwillu | top - 15:17:25 up 21:08, 2 users, load average: 6.95, 6.19, 5.06 | 22:17 |
| cwillu | Mem: 237068k total, 168120k used, 68948k free, 28k buffers | 22:17 |
| cwillu | Swap: 0k total, 0k used, 0k free, 22116k cached | 22:17 |
| cwillu | _lots_ of free memory | 22:17 |
| rcn-ee_lpt | yet it's stuck waiting for something.. | 22:18 |
Generated by irclog2html.py 2.7 by Marius Gedminas - find it at mg.pov.lt!