=== nchambers is now known as uplime === chat is now known as Guest84732 [01:02] Why does my wordpress images send so many request header cookies? Requesting one .png has like 16 cookies with it. whyyy === Guest20946 is now known as mgagne [01:55] Hey guys, I'm running Ubuntu on 16.04.1 and I have mod-security enabled. I'm trying to send a GET / POST request without a User Agent attached, but I can't, it causes a 403, this is the security notice: [msg "Request Missing a User Agent Header"] [severity "NOTICE"], is it possible to turn User Agent filtering off but leave the rest intact? [01:59] if this is documenting the same mod_security, there's a dozen instances of User-Agent on this page https://github.com/SpiderLabs/ModSecurity/wiki/Reference-Manual#REQUEST_HEADERS [03:50] has anyone used maas [05:37] jvwjgames: yes [06:32] Hello all, I am doing a preseeded installation of ubuntu 16.04.3 from a custom iso. - But the installer is failing on "apt-install or in-target is already running..." messages appear in syslog. - I have found a bug on lunchpad (1450597) however it seems to reference booting from usb stick. Bug is from 2013. The installation is made on a virtualization platform. [06:32] The preseed file itself works pxe-booted. [06:34] KERNEL /casper/vmlinuz.efi [06:34] ups [06:34] https://pastebin.com/7sJL846u <-- this is the append to kernel boot I am using. - That paste is from another isolinux.cfg I am working on parallell I am aware it is desktop when its casper. [07:38] version niceness - on update just "Unpacking python-html5lib (0.999999999-1)" passed by - really :-) ? [07:40] rmadison on that looks funny http://paste.ubuntu.com/p/V3rcYm77DK/ [07:40] seems to try to become a check for the max length of a version [07:40] * cpaelzer stops having fun now :-) === Nafallo_ is now known as Nafallo [13:59] jamespage: woo, zero fails on queens-proposed [13:59] coreycb: awesome! [13:59] jamespage: want to promote to -updates and get an announcement out? [13:59] coreycb: I've been pushing everything through to updates :-) [14:00] cinder just went [14:00] jamespage: they are mostly no-ops at this point so i think that's fine [14:00] last bit I think [14:00] yeah that was my take [14:00] jamespage: i'll start a doc [14:02] jamespage: oh nova-lxd is still at rc1 [14:06] coreycb: hmmm [14:06] tinwood: do we have a release for nova-lxd yet? [14:11] jamespage: just giving a final look at horizon [14:11] coreycb: always a good idea [15:04] jamespage: seeing 2 issues with launching an instance. network doesn't show up as a selection and admin doesn't have a Launch Instance button. [15:04] jamespage: on horizon, cli is fine [16:02] coreycb: jamespage: do you happen to have a sample for a git-build-recipe that builds for xenial based on UCA? would be great to have launchpad take over my builds ;) [16:14] frickler: hmm [16:14] frickler: not per say [16:14] coreycb: ^^ ? [16:16] frickler: i don't, sorry === smoser1 is now known as smoser [17:18] when a process uses too much RAM, and the os/kernel/whatever kills it - does anyone know what signal it uses to kill that process? (is it something i could catch)? [17:20] oom killer? [17:20] i am sure its SIGKILL. [17:26] yeah, it's SIGKILL [17:27] :/ of course it is, heh [17:28] trying to write a script to handle some rather cpu/memory intensive stuff and distribute it across a bunch of machines, but i'm worried the memory use will mark it for death from the oom killer [17:28] and not sure how to handle that situation [17:28] i don't want the handler at the end which brings all the results together to wait forever [17:30] you can tune the oom_score for a given process [17:30] sdeziel: it's a type of process where - if left alone without a time limit - *will* run out of memory [17:31] mordof: I'm not sure about the "wait forever" part? Are you saying the handler would wait for a dead process? [17:31] so changing score will just change the time frame it'll take before it dies [17:31] i've got a central process which dispatches each of the tasks - and then waits for the results through redis or some other medium [17:31] i could kill it after several hours if nothing came through.. but.. that's not very nice [17:32] could the workers send a heart beat every so often? [17:32] if a worker doesn't report inside 3 heart beat time, you conclude it's dead [17:33] sdeziel: sometimes i wonder why my brain just doesn't come up with these things... i've done that in the past, why that didn't occur to me... lol [17:33] or even something simpler: just note the PID of the worker and when that PID is gone, conclude the process was reaped by OOM [17:33] true [17:36] sdeziel: heartbeat will be a bit easier to implement here i think - gonna go that route, and then tune parameters for performance as needed. thanks [17:36] mordof: glad I could help [18:07] dpb1: powersj: rbasak: fyi, it looks like there might be some significant chnages in snapcraft that will let us trim our snap back down [18:07] not an immediate priority, but eventually [18:07] sweet [18:07] can also bundle the .pyc files [18:12] dpb1: yeah, i think we can actually trim a lot out, but i haven't had the cycles to figure out what yet [18:12] backlog [18:12] ya [18:13] our CI will help us make sure we don't break things in general, i hope === uplime is now known as nchambers === y0sh_ is now known as y0sh [21:50] jamespage: queens release announcement sent! [22:50] coreycb: nice [22:50] :) === z0_ is now known as z0