cjwatson | juliank: That's not a GCC bug; the ppc64el builders have known occasional guest memory corruption. | 00:33 |
---|---|---|
cjwatson | It's just occasional enough that we can live with it. | 00:33 |
infinity | cjwatson: I have high hopes that when they're xenial-on-xenial, that'll go away. | 00:35 |
infinity | (but I'm not holding my breath) | 00:35 |
cjwatson | Nor am I. What's the host right now, wily? | 00:35 |
cjwatson | The guests are xenial now. | 00:35 |
infinity | trusty + wily + cloud archive qemu, I think. | 00:35 |
infinity | Something like that. | 00:35 |
cjwatson | Ah yes, that assembly. | 00:36 |
infinity | Maybe trusty + vivid + cloud. | 00:36 |
cjwatson | I think it's a slightly-patched backport of the cloud qemu. | 00:36 |
infinity | cjwatson: The bit that irks me is that it doesn't happen on my ancient Frobisher-based builders. | 00:39 |
infinity | cjwatson: So either it's a qemu/kernel regression, or some patch in that build never made it upstream. | 00:39 |
infinity | juliank: Either way, it's so hard to reproduce that debugging it seems like a fantasy. | 00:39 |
infinity | Err. | 00:39 |
infinity | cjwatson: ^ | 00:39 |
infinity | (Or, I suppose it could be POWER8-specific, but that seems far-fetched) | 00:40 |
cjwatson | Yup. | 00:44 |
cjwatson | I've never bothered to collect actual figures, but IIRC it's something like one build in a hundren. | 00:45 |
cjwatson | *hundred | 00:45 |
cjwatson | Could be load-dependent for all I know. | 00:45 |
=== juliank_ is now known as juliank | ||
=== fginther` is now known as fginther | ||
masterchef | hey gang! I am interested in getting started in Ubuntu packaging and wanted to know if there were any good resources to consult | 04:21 |
tsimonq2 | masterchef: http://packaging.ubuntu.com/html/ | 04:24 |
tsimonq2 | it would also be worth looking at the archive processes | 04:25 |
tsimonq2 | so like SRUs and Proposed Migration | 04:25 |
tsimonq2 | if you know the archive, you know where you can solve problems | 04:26 |
stgraber | pitti: "lxc config set container environment.KEY value" will cause it to be set in init's environment | 08:50 |
stgraber | pitti: for arguments you could do "lxc config set container raw.lxc 'lxc.init_cmd=/sbin/init --debug'" | 08:50 |
=== DzAirmaX_ is now known as DzAirmaX | ||
=== JanC is now known as Guest3715 | ||
=== JanC_ is now known as JanC |
Generated by irclog2html.py 2.7 by Marius Gedminas - find it at mg.pov.lt!