=== jfarschman is now known as MilesDenver | ||
onicrom | woo 1.6 | 00:47 |
---|---|---|
onicrom | congrats sirs | 00:47 |
onicrom | and ladies | 00:47 |
bigjools | thanks onicrom | 01:24 |
bigjools | 1.6.1 hot on its heels :) | 01:24 |
bigjools | and 1.7 beta soon to follow | 01:25 |
onicrom | i had to rebuild my maas server | 01:25 |
onicrom | so im going to upgrade :) asap | 01:25 |
onicrom | building a zfs storage server and needed to replace the 'raid' cards with non-raid.... and my bios sucks it will only boot from the first 'drive' it sees | 01:27 |
onicrom | been through hell and back i tell you | 01:28 |
onicrom | ive seen some shit | 01:28 |
bigjools | sounds... fun | 01:28 |
onicrom | here is an example... i had to flash non raid firmware.. the flash tool only works in windows... i only have remote access to these machines and via kvm, which has a usb drive plugged in which is shareable if the kvm is on the computer yuoure using.. (USB1 speed) | 01:30 |
onicrom | so i found a winpe iso which i had to hack unetbootin to use as it required ntfs, which i then also had to hack syslinux to boot it... then i flashed the firmware ... yay... installed ubunut... yay... boot.. no love | 01:31 |
onicrom | takes 20mins to write the iso to usb, and 30mins to boot the damn thing (usb1 wooo) | 01:32 |
jtv | This. This bit of code I just ran into from some other people's project: | 05:37 |
jtv | pbp = bp;/* jwg, 9/6/96 - nice catch! */ | 05:37 |
jtv | This is why clear code matters so much. | 05:38 |
jtv | Also, non-ambiguous date formats. | 05:39 |
=== CyberJacob|Away is now known as CyberJacob | ||
bigjools | jtv: so the answer here, then, is to stop looking at other people's projects | 06:53 |
jtv | bigjools: ah yes, of course. | 06:56 |
bigjools | jtv: I'm a bit worried that your https://code.launchpad.net/~jtv/maas/allocate-ips-on-multiple-cluster-interfaces/+merge/231368 introduced race conditions | 07:36 |
jtv | Aren't we running in serializable nowadays? | 07:36 |
bigjools | no | 07:36 |
jtv | ! | 07:37 |
jtv | How did that change? | 07:37 |
bigjools | which is why in another branch Gavin did a load of fixes to cope with READ_COMMITTED | 07:37 |
jtv | /o\ | 07:37 |
bigjools | serializable didn't work so it was backed out | 07:37 |
bigjools | something broke in CI | 07:37 |
bigjools | can't remember what | 07:37 |
jtv | But at the time (Austin) we did that to solve a real production problem. | 07:37 |
bigjools | NFI | 07:38 |
bigjools | the new claim_static_ips you did looks very racy :( | 07:38 |
jtv | All Django code is racy. | 07:39 |
bigjools | well, yes :) | 07:39 |
bigjools | let me point you at what Gavin did, one sec | 07:39 |
bigjools | and hilariously, you reviewed it :) | 07:40 |
bigjools | https://code.launchpad.net/~allenap/maas/static-ip-address-allocate-fixes/+merge/231048 | 07:40 |
bigjools | I have to go, back for call | 07:41 |
=== CyberJacob is now known as CyberJacob|Away | ||
allenap | bigjools, jtv: We ran serializable in development but the change never made it into production configurations, iirc. | 08:33 |
dimitern | allenap, bigjools, jtv, rvba, Hey guys! A quick question re 1.6 and ip allocation for lxc containers - will the bridging setup still work (i.e. br0 as primary NIC on the node, eth0 as one of the ports) for containers to automatically get a DHCP-assigned IP? Or now we have to use the explicit IP reservation API? | 09:53 |
jtv | dimitern: we intend to keep it working with dhcp. | 09:54 |
dimitern | jtv, ok, that's great for short-term backwards compatibility | 09:55 |
jtv | If you want to though, you can still allocate a static IP through the API. But you'll have to configure the container's networking setup to use it. | 09:55 |
dimitern | we will implement explicit allocation at some point soon, but I'm glad I don't have to worry about it just yet :) | 09:55 |
dimitern | right | 09:56 |
jtv | dimitern: dhcp is served in the cluster interface's "dynamic IP range." | 10:01 |
jtv | If you allocate an IP through the API, you get one from the "static range." | 10:01 |
dimitern | jtv, right, but as far as container addressability is concerned, both ranges are equally fine to use from juju's perspective | 10:05 |
jtv | Right ho then. :) | 10:08 |
cjohnston | Mornin.. is anyone able to help me with bug #1279880? I seem to be having the same issue, however, for me, disabling the default network didn't help | 12:01 |
ubot5 | bug 1279880 in MAAS "Pserv responds to router IP instead of Requester in KVM virtual network" [Undecided,Invalid] https://launchpad.net/bugs/1279880 | 12:01 |
blake_r | rvba: could you review the backwards compatable api we talked about for boot sources | 13:29 |
blake_r | rvba: https://code.launchpad.net/~blake-rouse/maas/backward-compatible-boot-source-api/+merge/231499 | 13:29 |
magicrobotmonkey | Hi I have a cluster controller running on a node with a bonded interface | 14:15 |
magicrobotmonkey | and I can't get it to respond to dhcp requests from other nodes | 14:15 |
=== roadmr is now known as roadmr_afk | ||
magicrobotmonkey | I'm getting a lot of "Duplicate ACK for blocknum X" in pserv.log | 15:52 |
=== roadmr_afk is now known as roadmr | ||
lazyPower | Greetings, question about my maas configuration. What do i need to stuff on my cluster controller to fix the 90curtain-aptproxy configuration that seems to be blocking any and all PPA action? | 18:16 |
lazyPower | i do a ton of charm testing on a vmaas setup, and this triggers more false positives than anything - i wind up remoting into the unit and commenting out the 90curtain-aptproxy 100% of the time when a charm uses a ppa. | 18:17 |
lazyPower | is theproper method here to install a squid-deb-proxy on my controller node, or do i hack on a partial somewhere? | 18:25 |
=== CyberJacob|Away is now known as CyberJacob | ||
lazyPower | and instaling squid-deb-proxy didn't seem to make a difference. | 20:42 |
lazyPower | bummer :( | 20:42 |
=== CyberJacob is now known as CyberJacob|Away | ||
=== CyberJacob|Away is now known as CyberJacob |
Generated by irclog2html.py 2.7 by Marius Gedminas - find it at mg.pov.lt!