| *** pdar has quit IRC | 02:22 | |
| *** pdar has joined #baserock | 02:22 | |
| *** rdale has quit IRC | 02:34 | |
| *** bwh has quit IRC | 03:06 | |
| *** bjdooks has quit IRC | 03:06 | |
| *** bwh has joined #baserock | 03:13 | |
| *** bjdooks has joined #baserock | 03:13 | |
| *** mauricemoss_ has quit IRC | 07:10 | |
| *** mauricemoss_ has joined #baserock | 07:10 | |
| *** a1exhughe5 has joined #baserock | 07:12 | |
| *** Albert has joined #baserock | 07:15 | |
| *** mdizzle has joined #baserock | 07:37 | |
| *** mdunford has joined #baserock | 07:54 | |
| *** jonathanmaw has joined #baserock | 07:54 | |
| Kinnison | paulsherwood: Morning, and havva-PR: https://github.com/devcurmudgeon/ybd/pull/19 | 07:58 |
|---|---|---|
| *** bashrc has joined #baserock | 08:09 | |
| *** mike has joined #baserock | 08:10 | |
| *** mike is now known as Guest86202 | 08:10 | |
| *** edcragg has joined #baserock | 08:26 | |
| *** rdale has joined #baserock | 08:27 | |
| *** ssam2 has joined #baserock | 08:44 | |
| *** ChanServ sets mode: +v ssam2 | 08:44 | |
| *** franred has joined #baserock | 08:45 | |
| paulsherwood | Kinnison: thanks :-) | 08:51 |
| jjardon | Hi, Id like to check the config.log of one of the chunks of one of the systems that are being built in the ci: is that info available somewhere? | 08:56 |
| * SotK tries to work out what the test at the bottom of morph's "morphless-chunks" test actually does | 08:57 | |
| SotK | I guess its just checking that the artifact is a tarball | 08:57 |
| ssam2 | jjardon: the config.log is thrown away after the build completes, along with the rest of the staging area | 08:57 |
| ssam2 | jjardon: the .build-log file (stdout and stderr of the build) is available, which might be enough | 08:58 |
| ssam2 | jjardon: there's no easy way to get at it, though. I really want a `morph show-build-log` command | 08:58 |
| jjardon | ssam2: URL? | 08:58 |
| ssam2 | let me get back to you | 08:59 |
| ssam2 | jjardon: if you know the cache-key of the artifact, you can get the build log with `curl --get http://cache.baserock.org:8080/1.0/artifacts?filename=a68c2fc3fa2513b5b0717dd256e092f74b1b42f0fb71055b80aa605d682d5c50.build-log` | 09:02 |
| ssam2 | where the long hex string is the cache key | 09:03 |
| ssam2 | `morph list-artifacts` can help you find the cache key | 09:03 |
| jjardon | ssam2: thank you! | 09:04 |
| *** mdunford has quit IRC | 09:06 | |
| *** wschaller has joined #baserock | 09:07 | |
| *** gary_perkins has joined #baserock | 09:08 | |
| pedroalvarez | Q: as part of the openstack integration work we have generated systemd units for postgres, rabbitmq, openvswitch and apaache-httpd. Would it be ok if we install them in their chunks so that they are not openstack-sytem specific? | 09:08 |
| ssam2 | I think the original ideal of systemd was that upstreams would ship unit files that worked everywhere | 09:09 |
| ssam2 | i'm in favour of working towards that goal | 09:09 |
| pedroalvarez | ssam2: makes sense | 09:10 |
| richard_maw | inlining in the post-install-commands makes the most sense to me | 09:10 |
| pedroalvarez | yeah, that too | 09:10 |
| jjardon | I think I found a bug in our dbus setup: seems one of the binaries is not installed with the correct permissions: http://paste.baserock.org/niwopupuxe | 09:12 |
| jjardon | not sure why this is happening as we are running as root (so id -u should be 0)? | 09:13 |
| radiofree | is there any documentation for things like post-install-commands (is there a pre-install-command? post-build-commands? etc..) anywhere? | 09:13 |
| *** bfletcher has joined #baserock | 09:13 | |
| straycat | it used to be in the README, it should now be on the wiki | 09:13 |
| radiofree | it's in the readme, thanks | 09:14 |
| richard_maw | jjardon: in what context is this script being run? | 09:14 |
| radiofree | maybe a guide to writing chunks would be useful on the wiki | 09:14 |
| ssam2 | radiofree: http://wiki.baserock.org/definitions/current | 09:15 |
| jjardon | richard_maw: dbus installation: http://git.baserock.org/cgi-bin/cgit.cgi/delta/dbus.git/tree/bus/Makefile.am?h=baserock/dbus-1.8.16-1-g3c05557/genivi/J-1.0#n224 | 09:15 |
| radiofree | ssam2: ah, ok, i failed to find that page! post-install-command didn't find anything | 09:15 |
| *** lachlanmackenzie has joined #baserock | 09:16 | |
| richard_maw | jjardon: ah, that's because our biuld sandboxing doesn't allow chmod/chown | 09:16 |
| richard_maw | being root is not the same as having all the capabilities | 09:16 |
| *** Krin has joined #baserock | 09:21 | |
| *** mdunford has joined #baserock | 09:23 | |
| jjardon | richard_maw: ok, so is it a bug in the Makefile.am, or should we set those manually in our dbus morphology? | 09:25 |
| richard_maw | set what manually? | 09:25 |
| radiofree | i already have the dbus untils for the systembus on a baserock system, if that's what you're refering to? | 09:26 |
| radiofree | s/untils/units | 09:26 |
| jjardon | radiofree: no, the permission and group of the dbus-daemon-launch-helper binary are not correct | 09:29 |
| radiofree | ah | 09:30 |
| jjardon | richard_maw: I meant change manually the group/permission of the dbus-daemon-launch-helper binary in some way | 09:31 |
| richard_maw | jjardon: in the chunk definition you'll run into the same problem as Makefile.am | 09:33 |
| richard_maw | jjardon: currently I think you'd need to set it at deployment time or first-boot | 09:33 |
| ssam2 | does that mean we have no setuid binaries built in Baserock, and can't have any? | 09:34 |
| Kinnison | if that were true, we'd have issues | 09:34 |
| jjardon | seems that the case | 09:34 |
| richard_maw | hm, ping traditionally needs to be setuid | 09:40 |
| richard_maw | I'm not 100% sure if it's the case though, since we _do_ use sudo | 09:40 |
| richard_maw | which has to be setuid | 09:40 |
| *** Krin has quit IRC | 09:46 | |
| paulsherwood | what system does mason build? | 09:46 |
| SotK | the systems defined in clusters/ci.morph | 09:47 |
| SotK | s/defined/listed/ | 09:47 |
| SotK | (and also only the ones which are of the same architecture as the Mason) | 09:48 |
| paulsherwood | thanks SotK | 09:49 |
| jjardon | richard_maw: yeah, seems the setuid is set correctly (I see rwsr-x--- here), the group is still incorrect though (should be messagedbus instead root) | 09:50 |
| paulsherwood | jonathanmaw: are you still subscribed to genivi-dev? | 09:51 |
| jonathanmaw | paulsherwood, I think so. | 09:51 |
| *** Krin has joined #baserock | 09:51 | |
| jonathanmaw | I see that you jumped into discussion on the topic "SAT feedback on the Tools Team Charter" | 09:51 |
| paulsherwood | jonathanmaw: yes, and then jeremiah started a new thread relative to gdp - i wonder if you would be willing to comment on that, please? | 09:53 |
| paulsherwood | (back to him and that list, i mean) | 09:53 |
| paulsherwood | sorry to note that it's a private list, folks.... we're hoping that will change in future | 09:54 |
| ssam2 | Mason is fail: https://mason-x86-64.baserock.org/log/fe2761b10c06ac2e213fa339560100a6158c2bc5--2015-04-14%2009:38:24.log | 09:55 |
| ssam2 | seems to be Mesa failing due to -Werror | 09:55 |
| * paulsherwood is ready to +2 a patch? | 09:56 | |
| ssam2 | not sure what has changed, Mesa certainly hasn;t | 09:56 |
| jjardon | mmm, there were not recent changes in mesa or gcc | 09:57 |
| *** ctgriffiths_ has joined #baserock | 09:57 | |
| ssam2 | seems to be the D-Bus marshalling patch that broke things, but that doesn't really make sense | 09:57 |
| ssam2 | no, it was the previous commit. 'Add openstack ansible modules to ansible.morph'. right. | 09:58 |
| jjardon | this is suspicious: Progress: Ready to build mesa-misc: waiting for a worker to become available | 09:58 |
| ssam2 | ah... systemd update | 09:58 |
| ssam2 | the error is from something that links against udev | 09:58 |
| ssam2 | or at least, in a function called 'udev_device_new_from_fd' | 09:59 |
| ssam2 | i'll try a patch that disables -Werror in Mesa, don't have time to dig further | 10:01 |
| persia | Can we unupdate systemd instead? | 10:01 |
| ssam2 | seems it doesn't have -Werror, it specifically enables '-Werror=implicit-function-declaration -Werror=missing-prototypes' | 10:01 |
| ssam2 | so it might go deeper | 10:01 |
| ssam2 | persia: I believe it was updated for a reason | 10:02 |
| ssam2 | i don't know more than that | 10:02 |
| persia | We don't have pre-merge CI yet, but if we revert patches that break things, we stand a better chance of emulating it. | 10:02 |
| * paulsherwood is bisecting, since we don't know for sure which patch broke this | 10:02 | |
| *** ctgriffiths has quit IRC | 10:02 | |
| persia | Sure, but the person updating it broke things for everyone else, so needs to do it again, I'd think. | 10:02 |
| paulsherwood | yup. and since i did one of the patches which *might* have caused it, i'm bisecting :) | 10:03 |
| paulsherwood | or we could revert back to last-known-good? | 10:03 |
| pedroalvarez | I'd say only reverting the commit that breaks this | 10:05 |
| pedroalvarez | but, if this is because the systemd upgrade, then the team which is integraing openstack in baserock will be very sad | 10:05 |
| persia | pedroalvarez: No easy way to do that with git without breaking history. Revert-branch-to-parent is safe. Reparenting existing commits is unsafe. | 10:06 |
| franred | sad = we will have problems using neutron....loads | 10:06 |
| pedroalvarez | franred: openvirtualswitch iirc | 10:06 |
| persia | franred: neutron is why systemd updated? then neutron apparently has an interesting interaction with mesa that needs sorting :) | 10:06 |
| * persia expects neither the neutron nor the mesa developers expected this | 10:07 | |
| jjardon | Ah, seems this has already detected in gnome-continuous: http://lists.freedesktop.org/archives/mesa-dev/2015-March/079104.html . Fixed in mesa 771cd266b9d00bdcf2cf7acaa3c8363c358d7478 | 10:07 |
| pedroalvarez | jjardon: great | 10:07 |
| ratmice__ | so is the problem that CAP_FOWNER gets dropped, so root can no longer mess around with permissions of non-root users? | 10:08 |
| persia | jjardon: So we just update the mesa SHA1, and all should be fine? | 10:10 |
| pedroalvarez | I think there were problems re. upgrading mesa (for x86?) | 10:10 |
| pedroalvarez | I may be wrong | 10:10 |
| pedroalvarez | if that is the case, I'd cherrypick the fix in a branch | 10:11 |
| jjardon | yeah, we cannot upgrade mesa at the moment, we can cherry pick it though | 10:11 |
| jjardon | but seems out mesa mirror is not up-to-date: http://git.baserock.org/cgi-bin/cgit.cgi/delta/mesa.git/ | 10:11 |
| pedroalvarez | jjardon: oh, i'll take a look at it soon and poke you once it's working again | 10:12 |
| jjardon | Do we change something so #ifdef HAVE_SYSFS will fail? | 10:14 |
| Kinnison | richard_maw: system integration commands get to modify *anywhere* in the system root, yes? | 10:15 |
| richard_maw | Kinnison: I believe so | 10:16 |
| Kinnison | richard_maw: Okay, this bit of ybd hacking will be "fun" :) | 10:16 |
| Kinnison | richard_maw: thanks | 10:16 |
| *** Albert has quit IRC | 10:19 | |
| *** Albert has joined #baserock | 10:19 | |
| richard_maw | it's the slowest bit of building currently, as we don't have SotK's work on making it do copy on write for that rootfs | 10:21 |
| Kinnison | nod. | 10:22 |
| Kinnison | ybd will need specialcasing for systems currently anyway | 10:22 |
| * paulsherwood sighs :) | 10:22 | |
| Kinnison | paulsherwood: I have plans | 10:22 |
| Kinnison | paulsherwood: and plans on plans | 10:22 |
| paulsherwood | cool :-) | 10:22 |
| Kinnison | paulsherwood: but many are still nascent squirmy things which could be stepped on if I'm not careful | 10:23 |
| Kinnison | so I hide them away in my plan pouch | 10:23 |
| jonathanmaw | When I was building the automotive message broker, it checked for the json-c package under the name "json", and tried to include the json-c header as <json/json.h> (which I had to change to <json-c/json.h> to get it to build) | 10:24 |
| jonathanmaw | was it expecting something non-standard? | 10:24 |
| Kinnison | jonathanmaw: the NetSurf project has had similar issues in the past | 10:24 |
| Kinnison | jonathanmaw: json-c changed their name/paths at some point | 10:24 |
| nowster | Indeed! We had problems with that on one of the projects here. | 10:33 |
| nowster | Lots of the code assumed the old location. | 10:33 |
| paulsherwood | jonathanmaw: submit a patch to amb ? | 10:34 |
| jonathanmaw | paulsherwood: I'll add that to my to-do list | 10:34 |
| paulsherwood | heh :) | 10:35 |
| pedroalvarez | ok, the weirdest error I've ever seen in the lorry-controller: http://paste.baserock.org/okamudakix | 10:43 |
| richard_maw | gc already running sounds like the repository got stuck at one point and didn't clean up its state which said a gc was running | 10:45 |
| pedroalvarez | "Auto packing the repository in background for optimum performance." | 10:47 |
| pedroalvarez | so, git is doing this itself before we tell it to do it | 10:48 |
| Kinnison | Must be a new git behaviour | 10:49 |
| bjdooks | git has has auto-compaction for some time. | 10:50 |
| pedroalvarez | git version 2.3.0 | 10:50 |
| Kinnison | bjdooks: usually only on commit | 10:52 |
| Kinnison | bjdooks: since pull typically only creates packs not loose files | 10:52 |
| persia | Does that remove unanchored refs? If so, we might want to try to cause that to be disabled for lorries. | 10:52 |
| Kinnison | it may remove unanchored commits | 10:52 |
| Kinnison | but only from lorry's copy, not the copy in the git server which baserock then uses | 10:53 |
| pedroalvarez | indeed | 10:53 |
| pedroalvarez | so, if git autopacks automatically, should we remove the `git gc` in the lorrying process? | 10:54 |
| pedroalvarez | I'd prefer if we can disable the autopacking | 10:55 |
| Kinnison | It might be that the new autopacking is more efficient than the explicit gc we did | 10:55 |
| pedroalvarez | true | 10:57 |
| Kinnison | richard_maw: You mentioned WIP for copy-on-write for system artifact construction -- do you happen to have a link to that? | 11:00 |
| * Kinnison has a slow HDD and is wishing he hadn't written a copy installer now | 11:00 | |
| richard_maw | Kinnison: it's SotK's overlayfs patches | 11:00 |
| Kinnison | hmm | 11:02 |
| Kinnison | ta | 11:02 |
| * Kinnison makes a partition with more room | 11:02 | |
| Kinnison | 3G spare isn't enough for this work | 11:02 |
| pedroalvarez | jjardon: http://git.baserock.org/cgi-bin/cgit.cgi/delta/mesa.git/ | 11:06 |
| *** mdizzle is now known as mariaderidder | 11:24 | |
| *** bashrc has quit IRC | 11:31 | |
| ssam2 | I had missed that commit http://git.baserock.org/cgi-bin/cgit.cgi/baserock/baserock/definitions.git/commit/?id=3075e14d544fc14593c9ead8d1e9d44234135076 changes the Linux repo we use from delta/linux.git to delta/linux-stable.git | 11:32 |
| *** bashrc has joined #baserock | 11:33 | |
| ssam2 | which mirrors git://git.kernel.org/pub/scm/linux/kernel/git/stable/linux-stable.git instead of git://git.kernel.org/pub/scm/linux/kernel/git/torvalds/linux.git | 11:34 |
| jjardon | ssam2: yeah, linux repo doesnt contain the minor releases (3.19.2, for example) | 11:34 |
| ssam2 | right, makes sense | 11:34 |
| jjardon | pedroalvarez: thanks, patch coming! | 11:35 |
| pedroalvarez | cool! thanks! | 11:41 |
| *** wschaller has quit IRC | 12:09 | |
| *** fay_ has quit IRC | 12:40 | |
| *** edcragg has quit IRC | 12:41 | |
| *** flatmush has quit IRC | 12:41 | |
| *** petefoth has quit IRC | 12:41 | |
| *** nowster has quit IRC | 12:41 | |
| *** a1exhughe5 has quit IRC | 12:41 | |
| *** lachlanmackenzie has quit IRC | 12:41 | |
| *** sambishop has quit IRC | 12:41 | |
| *** mauricemoss_ has quit IRC | 12:41 | |
| *** mdunford has quit IRC | 12:41 | |
| *** tiagogomes_ has quit IRC | 12:41 | |
| *** tpollard_ has quit IRC | 12:41 | |
| *** gary_perkins has quit IRC | 12:41 | |
| *** bashrc_ has joined #baserock | 12:41 | |
| *** mariaderidder has quit IRC | 12:41 | |
| *** ssam2 has quit IRC | 12:41 | |
| *** bashrc has quit IRC | 12:41 | |
| *** Krin has quit IRC | 12:41 | |
| *** paulw has quit IRC | 12:41 | |
| *** franred has quit IRC | 12:41 | |
| *** Guest86202 has quit IRC | 12:41 | |
| *** jonathanmaw has quit IRC | 12:41 | |
| *** edcragg has joined #baserock | 12:41 | |
| *** mdunford has joined #baserock | 12:41 | |
| *** mauricemoss_ has joined #baserock | 12:41 | |
| *** tpollard_ has joined #baserock | 12:41 | |
| *** ssam2 has joined #baserock | 12:41 | |
| *** ChanServ sets mode: +v ssam2 | 12:41 | |
| *** Krin has joined #baserock | 12:41 | |
| *** franred has joined #baserock | 12:41 | |
| *** lachlanmackenzie has joined #baserock | 12:41 | |
| *** nowster has joined #baserock | 12:41 | |
| *** jonathanmaw has joined #baserock | 12:41 | |
| *** a1exhughe5 has joined #baserock | 12:41 | |
| *** paulw has joined #baserock | 12:41 | |
| *** sambishop has joined #baserock | 12:41 | |
| *** mariaderidder has joined #baserock | 12:41 | |
| *** Guest86202 has joined #baserock | 12:41 | |
| *** fay_ has joined #baserock | 12:41 | |
| *** gary_perkins has joined #baserock | 12:42 | |
| *** flatmush has joined #baserock | 12:42 | |
| *** tiagogomes has joined #baserock | 12:42 | |
| *** petefoth has joined #baserock | 12:42 | |
| paulsherwood | jonathanmaw: your email has been well received :) | 12:43 |
| pedroalvarez | I ended up sending a Yocto patch to the Genivi patch tracker last genivi release | 12:46 |
| * jonathanmaw also notes the problem with E-mail threads going to multiple mailing lists - my reply got sent to mailing lists that I'm not a member of, so they're pending approval :/ | 12:48 | |
| *** bashrc has joined #baserock | 12:58 | |
| paulsherwood | pedroalvarez: why on earth would you do that? what in the world were you thinking? :-) | 12:58 |
| pedroalvarez | heh. Well, I found that it wasn't building, I checked out why it was building for them, and then I realized that they were carrying a patch. Given that they did the patch it seemed unfair to me to re-do it and send it with my name on it | 13:02 |
| jjardon | can someone take a look to this bashrc patches (they already have a +1): https://gerrit.baserock.org/#/q/status:open+project:baserock/baserock/definitions+branch:master+topic:bashrc | 13:08 |
| bwh | jonathanmaw: My approach with closed mailman lists is to subscribe and then turn off mail | 13:08 |
| paulsherwood | :) | 13:09 |
| bashrc | first baserock patch merged | 13:13 |
| Kinnison | congrats | 13:15 |
| jjardon | bashrc: welcome! (and thanks) | 13:16 |
| bashrc | :) | 13:16 |
| straycat | :) | 13:18 |
| paulsherwood | w00t! | 13:21 |
| tlsa | why's the openssl chunk called "openssl-new"? | 13:40 |
| Kinnison | because it's a new source for the repo | 13:40 |
| Kinnison | IIRC | 13:40 |
| tlsa | right | 13:41 |
| *** wschaller has joined #baserock | 13:43 | |
| jjardon | pedroalvarez: I think http://paste.baserock.org/nucujajimo should be added to the instructions here? (in the 1 point): http://wiki.baserock.org/guides/how-to-cross-bootstrap/ | 13:46 |
| pedroalvarez | jjardon: yes, whatever is included in the bootstrap systems these days | 13:47 |
| jjardon | ok, will update the page | 13:48 |
| pedroalvarez | maybe we can point that out with a link to a bootstrapsystem in the master branch of definitions? | 13:48 |
| jjardon | pedroalvarez: yep, I was thinking about that | 13:48 |
| pedroalvarez | jjardon: thanks for fixing it :) | 13:49 |
| jjardon | ssam2: I can not authenticate with my google account in the wiki, known bug? | 13:49 |
| pedroalvarez | jjardon: so I take that you are going to bootstrap to a new architecutre? | 13:49 |
| pedroalvarez | jjardon: you can use the openid to authenticate in the meantime | 13:50 |
| jjardon | ssam2: the error is: "OpenID auth request contains an unregistered domain: http://wiki.baserock.org/ikiwiki.cgi" | 13:50 |
| paulsherwood | mason still borked? | 13:59 |
| radiofree | ERROR: Insufficient space on disk | 14:00 |
| *** mdunford has quit IRC | 14:01 | |
| radiofree | i don't know, it's confusing, the log says that early on but the goes on to build loads of things | 14:01 |
| radiofree | "2015-04-14 13:51:54 Finished building weston-system-x86_64-generic-rootfs" | 14:01 |
| pedroalvarez | I just created some space in both mason | 14:05 |
| ssam2 | jjardon: no idea about that. Use https://openid.baserock.org/ | 14:05 |
| ssam2 | or contact Branchable and ask what's up | 14:05 |
| *** mdunford has joined #baserock | 14:16 | |
| jonathanmaw | ooh, the AudioManager PoC is going to appear in a *real* repository this week, quite possibly adhering to FHS, as well. | 14:19 |
| paulsherwood | jonathanmaw: great news. i wonder if they were already working towards that? | 14:20 |
| jonathanmaw | paulsherwood: quite probably, since it comes as a response to the code review, not to me sticking my oar in. | 14:21 |
| paulsherwood | :-) | 14:21 |
| jonathanmaw | how to integrate the FSA is under discussion, and apparently yocto's postponed integrating it until after the AMM, since there's been discussion on how to integrate it into the LayerManager | 14:23 |
| radiofree | layermanager? weston surely? | 14:24 |
| jonathanmaw | radiofree: probably wayland-ivi-extension, since that implements the old LayerManager API | 14:25 |
| jonathanmaw | (more or less) | 14:25 |
| radiofree | yeah, it's confusing to refer to it as LayerManager though | 14:26 |
| * straycat did a little more digging and found http://support.ntp.org/bin/view/Support/LockingNtpdInMemory | 14:26 | |
| radiofree | though "weston-ivi-shell slash wayland-ivi-extension" hardly rolls off the tongue | 14:27 |
| straycat | so maybe should just set rlimit memlock 0 in the ntp.conf file? | 14:27 |
| straycat | (this will allow us to put hostnames back in the conf file) | 14:28 |
| straycat | thoughts? | 14:29 |
| *** bashrc has quit IRC | 14:35 | |
| *** bashrc_ has quit IRC | 14:35 | |
| jjardon | ssam2: np, I used my old launchpad account in the end :) | 14:35 |
| straycat | alright let's do it then | 14:35 |
| richard_maw | straycat: I'm a little IO starved right now, I'm still loading your link to see what's going on :-) | 14:36 |
| paulsherwood | straycat: who are you seeking feedback from? i'd comment but i don't understand anything at that level... | 14:36 |
| *** bashrc has joined #baserock | 14:37 | |
| straycat | paulsherwood, it's okay i just wanted to check with anyone who might be around | 14:38 |
| *** rdale has quit IRC | 14:38 | |
| *** rdale has joined #baserock | 14:39 | |
| richard_maw | straycat: So ntp segfaults if it's allowed to lock memory, and it needs to use memory for hostnames? | 14:41 |
| rjek | Wait, why is it using locked memory at all? Surely it has no secrets? | 14:42 |
| straycat | richard_maw, i'm not exactly sure it reports "out of memory" to stderr, i couldn't see what was going wrong from the strace | 14:42 |
| straycat | the call to mlockall succeeded | 14:42 |
| * rjek seems to recall that there's a sysconf for how much locked memory non-priviledged processes can have, and by default it is 64kB. | 14:44 | |
| straycat | http://sprunge.us/KXDS is the strace if it helps | 14:45 |
| rjek | it appears to exit in the middle of a syscall | 14:46 |
| ratmice__ | rjek: I imagine it has more to do with avoiding paging latency | 14:47 |
| richard_maw | rjek: threads are involved | 14:47 |
| rjek | ratmice__: Hmm, perhaps. But you'd hope it'd never get swapped as it'll routinely be hot. | 14:47 |
| rjek | straycat: strace -f? | 14:47 |
| straycat | richard_maw, in any case presumably the amount being reserved is not enough | 14:47 |
| richard_maw | could be | 14:48 |
| *** petefoth has quit IRC | 14:49 | |
| straycat | rjek, ahh thanks now it's quite obvious that the calls to mmap() fail | 14:50 |
| straycat | http://sprunge.us/Mjhh | 14:51 |
| rjek | is it obvious why, though? :) | 14:51 |
| * rjek learns what MAP_NORESERVE does and is sick a bit in his mouth | 14:52 | |
| straycat | not really, i would expect the call to mlockall to fail earlier on | 14:52 |
| rjek | 28388 setrlimit(RLIMIT_MEMLOCK, {rlim_cur=32768*1024, rlim_max=32768*1024}) = 0 | 14:53 |
| *** bashrc has quit IRC | 14:56 | |
| *** bashrc has joined #baserock | 14:57 | |
| straycat | ahh it uses MCL_FUTURE | 14:58 |
| straycat | okay that makes more sense | 14:58 |
| straycat | ratmice__, that's what i gather from http://support.ntp.org/bin/view/Support/LockingNtpdInMemory | 15:09 |
| straycat | so we could set the rlimit to something much higher? i've seen it fail with 128 though >.> so i was thinking of setting it to 256 | 15:11 |
| straycat | (we can set this in the ntp.conf) | 15:11 |
| * richard_maw just submitted https://gerrit.baserock.org/284 but hasn't got time to test it | 15:25 | |
| * rjek unclear how to +1 that | 15:26 | |
| straycat | https://bugzilla.redhat.com/attachment.cgi?id=474441 | 15:26 |
| rjek | straycat: I wonder if it works on 32 bit systems and not on 64 bit systems, due to larger structure sizes? | 15:27 |
| straycat | from https://bugzilla.redhat.com/show_bug.cgi?id=575874 | 15:27 |
| straycat | rjek, yes, and that's exactly the problem in that report ^ | 15:27 |
| straycat | weird that they patched the source rather than setting the rlimit in the config file? maybe ntpd didn't have support for that back in 2010 | 15:28 |
| rjek | straycat: man ntp.conf doesn't immediately reveal anything to me that woudl suggest you can do it from the config file? | 15:28 |
| Kinnison | rjek: I've +1d it | 15:28 |
| rjek | Although I'm on wheezy, which is ancient. | 15:28 |
| richard_maw | rjek: you need an account on the openid.baserock.org provider, to log into gerrit.baserock.org with that provider, and click the reply button and select the +1 radio button | 15:28 |
| richard_maw | (note, there's two interfaces, I don't recall if there's a reply button on the other interface, and which interface you get by default) | 15:29 |
| rjek | I can't use an arbitrary OpenID? | 15:29 |
| richard_maw | rjek: sadly no, gerrit doesn't support that, AIUI it needs to be configured with a fixed openid provider, rather defeating the purpose | 15:30 |
| richard_maw | and we didn't want to tie ourselves to the launchpad openid | 15:30 |
| * rjek facepalms | 15:30 | |
| ssam2 | Gerrit supports arbitrary openid's | 15:30 |
| * richard_maw is happy to be proven wrong | 15:30 | |
| Kinnison | ssam2: it does? | 15:30 |
| ssam2 | but it's trivial to set up an anonymous openID provider, so that would defeat the point of having an authentication mechinism | 15:30 |
| straycat | rjek, well http://support.ntp.org/bin/view/Support/LockingNtpdInMemory did say it's not well documented >.> | 15:31 |
| straycat | but you clearly can do it | 15:31 |
| * Kinnison keeps meaning to set up an openID provider which works by you asserting fooprovider/FULLGPGKEYFINGERPRINT/TYPE/SIZE and then it challenges you to sign something with your key | 15:31 | |
| straycat | anyway, we'll use 256 and hope that's enough | 15:31 |
| ssam2 | storyboard only supports OpenID single-signon, that's probably what you were thinking of richard_maw | 15:32 |
| rjek | Anyway, I've managed now :) | 15:33 |
| richard_maw | it still needs to be +2'd and merged, and I won't do that without testing | 15:48 |
| * rjek has nothing to try it on atm | 15:49 | |
| *** Krin has quit IRC | 15:56 | |
| * nowster has something locally that's intermittently interfering with 60kHz reception. | 15:57 | |
| jmacs | MSF? | 15:59 |
| nowster | yup | 16:03 |
| nowster | Bearing from Manchester would be about 20°W of N | 16:04 |
| nowster | (or 340°) | 16:04 |
| nowster | I've been toying with the idea of getting a DCF77 module, too. | 16:07 |
| rjek | I appear to have walked into a parellel universe where nothing is the same. | 16:07 |
| paulsherwood | rjek: that happens every day at codethink :) | 16:07 |
| rjek | heh | 16:08 |
| *** a1exhughe5 has quit IRC | 16:32 | |
| jjardon | any reason to have 2 separate bsp-x86_32-generic and bsp-x86_64-generic strata? they are identical | 16:34 |
| *** mariaderidder has quit IRC | 16:36 | |
| tlsa | jjardon: different chunk morphs | 16:43 |
| jjardon | tlsa: the chunks morphs are identical as well | 16:44 |
| jjardon | well, only the name changes (32/64) | 16:44 |
| ssam2 | would be good to merge them then I think | 16:46 |
| richard_maw | where would the definition live then? | 16:46 |
| jjardon | I already have a patch to name it strata/bsp-x86_both-generic.morph | 16:48 |
| ssam2 | please, just bsp-x86-generic :) | 16:48 |
| jjardon | 42 files changed, 369 insertions(+), 739 deletions(-) | 16:49 |
| jjardon | ssam2: ok | 16:49 |
| pedroalvarez | No changes between HEAD and gerrit/master. Submitting for review would | 16:49 |
| pedroalvarez | be pointless. | 16:49 |
| richard_maw | er what? | 16:50 |
| pedroalvarez | I hope I haven't pushed anything to master.. | 16:50 |
| richard_maw | doesn't look like it | 16:51 |
| pedroalvarez | nope... phew | 16:51 |
| paulsherwood | pedroalvarez: i get that issue, haven't got round to trying straycat's suggested fix | 16:51 |
| pedroalvarez | paulsherwood: oh, when was this fix suggested? | 16:52 |
| paulsherwood | git commit --amend, and remove change-id | 16:52 |
| pedroalvarez | ohh | 16:52 |
| pedroalvarez | ok | 16:52 |
| *** Albert has quit IRC | 16:53 | |
| paulsherwood | maybe with another git commit --amend after that? :) | 16:54 |
| * paulsherwood has no idea if/how this magic works, though | 16:54 | |
| pedroalvarez | it does | 16:55 |
| pedroalvarez | https://gerrit.baserock.org/#/q/status:open+project:baserock/baserock/definitions+branch:master+topic:baserock/openstack-in-baserock | 16:55 |
| ssam2 | anyone have a clue why `morph controller-daemon` might totally not listen on port 7878, even though it's running? | 16:55 |
| ssam2 | the log file clearly shows 'controller-initiator-port = 7878' in the config.. | 16:56 |
| *** jonathanmaw has quit IRC | 16:56 | |
| ssam2 | 'git master' of Morph | 16:56 |
| paulsherwood | pedroalvarez: some merge conflicts there | 16:57 |
| *** petefoth has joined #baserock | 16:57 | |
| paulsherwood | and lots red tones | 16:57 |
| richard_maw | paulsherwood: that's because gerrit is bad at patch series | 16:57 |
| richard_maw | it's conflicting with itself in a way which isn't a problem if you consider the previous patch to have been merged when you check for whether it conflicts | 16:57 |
| *** Guest86202 has quit IRC | 17:03 | |
| ssam2 | ah, seems my problem might be that some of the distbuild services had started, but not all | 17:04 |
| *** lachlanmackenzie has quit IRC | 17:05 | |
| *** bashrc has quit IRC | 17:06 | |
| *** Albert has joined #baserock | 17:07 | |
| jjardon | What the arch parameter un the system morphology do? | 17:12 |
| jjardon | I though the arch depends on the bsp you choose | 17:13 |
| *** mdunford has quit IRC | 17:14 | |
| pedroalvarez | jjardon: is only possible to set an arch in the system morphology | 17:14 |
| pedroalvarez | systems are per arch | 17:14 |
| paulsherwood | ybd ignores the arch field currently - i pass it in as a param on the command line :) | 17:15 |
| jjardon | Yeah, i know, that's why I asked :) | 17:15 |
| pedroalvarez | ahh | 17:16 |
| pedroalvarez | sorry, brain frazzled | 17:16 |
| ssam2 | morph sets TARGET and TARGET_STAGE1 env variables based on 'arch' | 17:16 |
| ssam2 | i think. so that affects what compiler the bootstrap builds | 17:17 |
| *** ssam2 has quit IRC | 17:17 | |
| jjardon | OK, thanks | 17:17 |
| * paulsherwood misunderstood the question, sorry | 17:33 | |
| *** Albert has quit IRC | 17:34 | |
| *** wschaller has quit IRC | 17:35 | |
| *** edcragg has quit IRC | 17:47 | |
| *** Albert has joined #baserock | 17:49 | |
| *** sambishop has quit IRC | 18:05 | |
| *** sambishop has joined #baserock | 18:05 | |
| *** gary_perkins has quit IRC | 19:13 | |
| *** rdale has quit IRC | 19:44 | |
| *** Albert has quit IRC | 21:15 | |
Generated by irclog2html.py 2.15.3 by Marius Gedminas - find it at mg.pov.lt!