IRC logs for #baserock for Thursday, 2014-12-11

*** genii [~quassel@ubuntu/member/genii] has quit [Read error: Connection reset by peer]00:04
*** zoli_ [~zoli_@linaro/zoli] has joined #baserock05:02
*** zoli_ [~zoli_@linaro/zoli] has quit [Remote host closed the connection]05:10
*** zoli_ [~zoli_@linaro/zoli] has joined #baserock05:10
*** mariaderidder [~maria@82-70-136-246.dsl.in-addr.zen.co.uk] has joined #baserock08:55
*** bashrc [~motters@82-70-136-246.dsl.in-addr.zen.co.uk] has joined #baserock08:59
*** mdunford [~marcdunfo@82-70-136-246.dsl.in-addr.zen.co.uk] has joined #baserock09:02
*** tiagogomes [~tiagogome@213.15.255.100] has joined #baserock09:06
*** ssam2 [~ssam2@82-70-136-246.dsl.in-addr.zen.co.uk] has joined #baserock09:15
Mode #baserock +v ssam2 by ChanServ09:15
*** franred [~franred@82-70-136-246.dsl.in-addr.zen.co.uk] has joined #baserock09:23
*** CTtpollard [~tom@82-70-136-246.dsl.in-addr.zen.co.uk] has joined #baserock09:58
*** Krin [~mikesmith@82-70-136-246.dsl.in-addr.zen.co.uk] has joined #baserock10:15
pedroalvarezHi guys! I've experienced some problems when upgrading a Baserock system, I don't know why yet but after the upgrade the system doesn't log anything in the systemd journal unty I do `systemctl restart systemd-journald`10:20
pedroalvarezI've seen this problem in 3 different VM's now10:20
Kinnisonpedroalvarez: Does it log if you reboot a second time?10:21
jmacspdar: Did your build finish?10:21
richard_mawif it's a very recent version of systemd then there's been work to make it log to /run initially, then move it to /var as soon as it becomes writable10:22
richard_mawpedroalvarez: check whether it's got new ConditionFoo= in the unit file?10:22
pdarjmacs: it did not :( Ran out of disk space. Started it again when I got in though10:22
pdarjmacs: Did you do a build?10:23
pedroalvarezthe unit was running before I restarted it, though10:23
pedroalvarezI know I'm not giving useful information, but I think that is important to know that this can happen10:23
jmacspdar: Yeah, failed fairly quickly with "xfs/xfs.h not found"10:23
pedroalvarezKinnison: if I reboot it, it stops loging again10:27
Kinnisonpedroalvarez: Okay, sounds like the thing richard_maw mentioned then10:27
pdarjmacs: When configuring ceph you need to pass the --without-libxfs flag. 10:27
pedroalvarezhm.. okay... I'll leave the system and see if it  starts logging10:29
jmacspdar: Sure, how do you do that with morph though?10:29
pdarjmacs: Or could lorry xfsprogs and add to the build depends. but I was going to do that after I got ceph to work10:29
pdarjmacs: If you go to the `/strata/ceph-service/ceph.morh` you can change the `configure-commands:` and after `./configure` add the flag I mentioned10:31
pedroalvarezactually, the last thing I see when doing `journalctl -f`  is from 3 days ago :(10:31
pdarjmacs: Like this: https://github.com/padrigali/definitions/blob/pd/strata/ceph-service/ceph.morph10:31
jmacspdar: Aha, was missing that from your branch10:31
pedroalvarezI wonder if it's writing in /var before it mounts the btrfs volume /var10:32
pedroalvarezlooks like it10:33
ssam2ouch10:33
ssam2good spot10:33
richard_mawhm, the journal commit thingy is part of the brand new 218 release10:36
richard_mawif we were to sort it out with that, I think we'd need to start with /[/systems/default/run] mounted read-only reorder fstab so that /var gets mounted before / is remounted, then have / be remounted rw from fstab10:37
richard_mawbut it's probably a good idea to ask in #systemd to see if anyone there has any better ideas10:38
richard_mawwe might be able to get away with dropping in a `After=var.mount` in /etc/systemd/system/journald.service.d/after-var.conf, but then we'd end up without logging until /var is mounted10:39
jmacsg++: internal compiler error: Killed (program cc1plus) O_O10:39
pedroalvarezjmacs: not RAM enough?10:40
Kinnisonjmacs: RAM is typically the culprit for that10:40
Kinnisonpedroalvarez: snappish!10:40
pedroalvarezrichard_maw: I'd solve it using "RequiresMountsFor=/var/log/journal" in the journald unit10:41
jmacsIt has the baserock-recommended 2GB. I'll try uppping it...10:42
richard_mawI still worry that it'd prevent early-boot logging10:42
persiajmacs: That's a lower-bound guideline for building simple code.  In practice, if you're doing lots of C++, you want at least 4GB, and if you have a system that can use it, 8 or more.10:43
persiaThe same is also true for large Java, Haskell, or similar systems.10:43
richard_mawI think we ought to focus some effort on the work radiofree did to have multiple partitions for the jetson, so we can make it a first-class thing and have swap configuration possible at deployment time.10:44
jmacspersia: Naturally, but asking for a 8GB VM is going to be an onerous requirement for a lot of our users.10:45
jmacsPerhaps we should be pushing the chroot option more.10:45
persiajmacs: Yes, which is why the wiki says "2GB" :)  Still, if you want to get real work done ...10:45
jmacs... ignore the instructions?10:45
persiaOr perhaps, expand upon them.10:46
persiaRecognise they are guidelines to get started.10:46
pedroalvarezrichard_maw: adding the condition seems to work10:46
* SotK has set up swap on his VM to be able to build things (slow > not at all) in the past10:46
persiarichard_maw: I worry about that, because swap can thrash, which can be annoying, depending on the backing store10:46
persia(swap in virtio with a large-RAM virthost is not slow enough to thrash in most cases)10:47
petefothpersia: would it be worth adding that information about memory to the appropriate place on w.b.o?10:47
persiapetefoth: Possibly, but it really depends on the application.  Most folk are fine with 2GB.10:48
persiaOne needs to be working with large applications in one of a narrow set of languages to need lots of RAM at compile time.10:48
petefothpersia: seems to me that is *very* useful information for Baserock users. 10:49
persiaPerhaps in the "troubleshooting builds" section, we could add the error, and provide a hint that more memory would be useful?10:49
* SotK thinks it should be mentioned in the same place as the 2GB recommendation too10:50
persiapetefoth: For affected users, yes.  For everyone else, it is extra words they don't want to read.  I'm happy for it to be on the wiki.  I just think it needs to be discoverable for folk who need it, and not get in the way for users doing C programming for embedded or Python/Ruby/Javascript programming for cloud.10:50
* Kinnison would like it as a sidebar near the 2GB recommendation10:50
* persia defers to everyone else10:50
petefothKinnison: where is the '2GB recommendation'?10:51
KinnisonIt might be helpful in the FAQ / troubleshooting sections too though10:51
jmacsSeveral places....10:51
* SotK wouldn't enjoy setting up a 2GB VM following instructions, having a weird build issue only to discover the instructions could have mentioned it earlier10:52
petefothjmacs: doyou have the full text of the error message ort it what you posted above?10:52
jmacsDepends which virtualization software you're using. It's in the setup instructions for each.10:52
jmacspetefoth: No, I had to shut down the VM to increase the RAM...10:52
petefothjmacs: OK - no problem. I'll put the text you supplied in 'guides/build-issue, and adds something to the setup vm pages10:53
petefothhttp://source.baserock.branchable.com/?p=source.git;a=commitdiff;h=c445c5ada5c0b8fb20b2792bebe066b35b063a4511:22
petefoth(and the typo fix in  http://source.baserock.branchable.com/?p=source.git;a=commitdiff;h=00bbe40ea6a36ae96c848d5bfa6c802ec1a5a2d0)11:22
petefothhttp://source.baserock.branchable.com/?p=source.git;a=commitdiff;h=930ef450ecfa3e1fdf9e8e0e133f80283196357e11:22
petefoth    11:22
jmacsLooks good petefoth 11:23
*** zoli__ [~zoli_@0x5e91887a.adsl.cybercity.dk] has joined #baserock11:24
*** zoli_ [~zoli_@linaro/zoli] has quit [Ping timeout: 258 seconds]11:24
*** rdale [~quassel@180.Red-83-55-113.dynamicIP.rima-tde.net] has joined #baserock11:29
*** rdale_ [~quassel@68.Red-83-55-115.dynamicIP.rima-tde.net] has quit [Ping timeout: 264 seconds]11:32
pedroalvarezOk, after doing some tests, I think that adding 'RequiresMountsFor=/var/log/journal' to systemd-journald doesn't prevent early-boot logging: http://paste.baserock.org/rufoxeyuri11:37
pedroalvarezI'm not sure about how to proceed, I could send a patch to fix this in baserock, but I don't want to create delta in systemd. 11:45
pedroalvarezSo I guess I only can go to #systemd and discuss this11:45
richard_mawprobably best to ask upstream about it then11:46
richard_mawif it's as trivial a fix as you say, they may apply it then and there, so we don't have an upstream, we'd just have to use an unreleased version of systemd11:47
richard_mawthere's not been enough stuff after v218 that it would be unstable yet11:48
richard_maw(it's been bug fixes in kdbus support, which we don't use yet)11:48
ssam2perryl, SotK: I'm declaring http://testgerrit.baserock.org:8080/ officially usable11:56
ssam2not everything is lorried yet but baserock/baserock/definitions is there11:56
ssam2let me know any problems11:56
ssam2sadly you can't use http://openid.baserock.org/ with it for some reason, no time to investigate why at the moment11:57
ssam2but stackoverflow openids work11:57
richard_mawI think I saw that google's planning to drop openid support in gerrit all together11:57
richard_mawhttp://code.google.com/p/gerrit/issues/detail?id=271511:58
pedroalvarezyeah, we never managed to get that working11:59
perrylssam2: am i right in thinking that anything firehose sends to refs/for/master should show up here?12:01
ssam2richard_maw: really? that just looks like someone complaining about Google not providing OpenIDs any more12:01
ssam2perryl: yes, if it 12:01
ssam2...pushes it to test gerrit12:01
ssam2perryl: the push URLs are a bit funny12:01
ssam2http://testgerrit.baserock.org:8080/#/admin/projects/baserock/baserock/definitions lists the push URL for definitions.git12:02
ssam2which for me is ssh://samthursfield@testgerrit.baserock.org:29418/baserock/baserock/definitions12:02
perryli can see a git clone http://testgerrit ... but no push URL12:03
ssam2perryl: you're probably not logged in12:03
ssam2perryl: to push, you need an account12:03
pedroalvarezso maybe firehose needs an account?12:03
SotKas will Mason12:03
ssam2firehose needs its own account, indeed12:03
ssam2as will Mason12:04
ssam2it's possible to set up accounts for bots that don't need OpenIDs etc12:04
pedroalvarezbut I guess only admins can do  that12:04
ssam2maybe for now you two can use personal accounts, and when we deploy the instances to the baserock.org cloud we can set up proper accounts for them12:04
franredpedroalvarez, ssam2, I think gerrit has a non-online users group12:05
ssam2if that's too hard, come see me now and I'll show you how its done12:05
ssam2franred: it does indeed12:05
ssam2i.e. show you how to create special users12:05
ssam2Gerrit seems quite CPU intensive ...12:27
paulsher1oodit's java.... :)12:27
* paulsher1ood notices he can't login to that with a googleid, but guesses that may be expected12:30
persiaJava is usually slow because of memory requirements rather than CPU.  I wonder what it is doing.12:30
ssam2paulsher1ood: google IDs are not supported, apparently, as https://code.google.com/p/gerrit/issues/detail?id=2715 bemoans12:32
paulsher1ooddo we have an id provider?12:32
ssam2yes, but it doesn't work with Gerrit for some reason12:32
ssam2I've not had time to debug why as yet12:33
paulsher1oodheh12:33
ssam2http://openid.baserock.org/ does work with wiki.baserock.org though (or did when I last tested it)12:33
perrylclicking sign in via google gives "400. That's an error." thanks google, i would never have guessed!12:33
paulsher1oodhow do you and others login, then?12:33
ssam2stackoverflow openID right now12:33
paulsher1oodaha12:33
ssam2launchpad openID is probably fine too12:33
ssam2actually, since we allow any valid OpenID server, you could probably use http://openid.aliz.es/12:34
persiaWhen I was last looking at OpenID, I read a rumor that Google was deprecating that in favour of something else, which may be related.12:35
perryli can confirm launchpad openID works12:35
ssam2hmm, seems http://openid.aliz.es/ doesn't forward you back to Gerrit, so it can't be used after all.12:35
persiahttps://developers.google.com/+/api/auth-migration#timetable12:36
persiaMost importantly "f your app authenticates users by any means other than the Google+ Sign-In button, we recommend you switch your app to Google+ Sign-In..."12:36
persia(which implies an anti-federation strategy)12:37
*** zoli__ [~zoli_@0x5e91887a.adsl.cybercity.dk] has quit [Remote host closed the connection]12:37
ssam2testgerrit seems to be rejecting my attempts to push to 'refs/for/master', claiming that it's a non-fast-forward merge12:41
ssam2it's clearly wrong, but I need to eat some lunch before I do any more debugging12:41
ssam2now the machine has frozen again. I think this will need upgrading to have 2 CPUs, too.12:43
petefothhttp://testgerrit.baserock.org:8080/#/admin/projects/baserock/baserock/morph only gitves me a clone usrl (http://testgerrit...) I'm logged in using my UbuntuOne openId. IS this because I don't have push access?12:52
cyndisi think google is planning to move to this https://en.wikipedia.org/wiki/FIDO_Alliance12:52
petefothI'd like to be ablke to push my morph doc changes to gerrit for review instead of the ML12:53
cyndisactually, looks like they already support FIDO U2F logins for google services12:54
*** zoli_ [~zoli_@linaro/zoli] has joined #baserock12:56
cyndisthough that's still only as a second factor..12:56
*** mariaderidder [~maria@82-70-136-246.dsl.in-addr.zen.co.uk] has quit [Quit: Ex-Chat]12:56
petefothAnd the gerriit UI doesn't dynamically resize - I get horizontal scroll bars in Firefox and Chromium unless my window is nearly full screen width :)12:59
* SotK gets that in OpenStack too13:12
persiaThe gerrit UI is widely decried as annoying, to the point that several folk have written other tools to avoid it entirely.  Of these, I like git-review and gertty.13:17
petefothpersia: thanks - I'll look at those. 13:18
petefothMore important however is the lack of push urls :( I'm flummoxed13:18
persiagit-review takes care of that for me, but I've never set up a git-review compatible repo, so there may be a bit of magic.13:19
Kinnisonpetefoth: hav eyou registered an ssh key with it?13:35
petefothKinnison: yes13:35
KinnisonHmm13:36
* Kinnison looks13:36
Kinnisonpetefoth: on the page which shows you the clone url13:39
Kinnisonclick on the 'SSH' which doesn't look like it'd be active13:40
Kinnisonthat'll give you an ssh url13:40
petefothKinnison: got it - thanks13:40
petefothSo I'll add that url for the gerrit remote and try a git pull to check its working13:42
petefothYippee!13:44
ssam2I finally managed to push a branch to gerrit without using `git review`: I had to manually add the Change-Id: field to the commit message, and push to refs/for/master/sam/test rather than refs/for/master13:45
ssam2btw, please don't use this Gerrit for stuff other than testing yet13:45
ssam2all data in it will be lost at some point13:45
* petefoth sees the change entitled 'Delete everything in a fit of madness.' and boggles :)13:45
ssam2like i said, it's for testing :)13:45
* pedroalvarez will -1 it13:46
petefothssam2: so I'll carry on submitting changes to the ML13:46
ssam2please do. I'll announce when there's a Gerrit ready for use13:46
ssam2feel free to push to gerrit as well though if you want to test it out13:47
SotKssam2: could you not do `git push origin HEAD:refs/for/master`?13:47
SotKAlso, if you get the clone command from "clone with commit-msg hook" rather than "clone" you get a hook which automatically adds the Change-Id field :)13:48
ssam2SotK: when I did that, I got told that it was a non-fast-forward merge13:54
ssam2SotK: ah, that's cool about the commit-msg hook13:55
SotKssam2: that is strange :/13:55
*** zoli_ [~zoli_@linaro/zoli] has quit [Remote host closed the connection]14:02
paulsher1oodssam2: 2014-12-10 21:48:39 [Build 204/204] [devel-system-x86_64-generic] Fetching to local cache: artifact devel-system-x86_64-generic-rootfs14:13
* paulsher1ood apologises to ssam2 for whining about morph over lunch, though14:15
ssam2you're right, seems there are rootfs artifacts in cache.baserock.org14:19
ssam2and 33GB of free space... i wonder how :)14:20
* pedroalvarez has been taking care of the disk space14:20
Zara_could someone point me to a good resource explaining the differences between similar-looking git and morph commands (eg: git/morph branch; git/morph checkout)?14:22
persiapedroalvarez: How?14:22
pedroalvarezremoving rootfs artifacts sometimes14:23
perrylZara_: with regards to git, git branch creates the branch, git checkout switches to look at that branch14:23
paulsher1oodZara_: no such resource exists. you have morph help, for example and git help?14:23
petefothZara_: I don't know of any such resource. 14:23
persiaZara_: git branch generally causes the creation of a new branch.  Morph branch is equivalent to `git clone` on definitions, due to some UI issues with morph.14:23
petefothZara_:  and /msg Kinnison Help!14:23
persiaZara_: git checkout generally causes the working tree to resemble the requested item.  Morph checkout is deprecated.14:23
pedroalvarezwhy deprecated?14:24
SotKwhen did we deprecate checkout?14:24
Zara_perryl: yeah, I read up on the git side of things yesterday and it made sense. :) 14:24
persiaZara_: morph push and pull are special-case commands that may be useful if you have used morph's git-fat wrapper, and otherwise can be ignored.14:24
persiapedroalvarez: I7ve been told, and seen others told not to use checkout in this channel because it is confusing.14:25
persiaIf we didn't formally deprecate it, we have at least been recommending against new users starting with it for the past 8-10 months.14:25
paulsher1oodpersia - i wish it was, but it isn't deprecated yet14:25
* pedroalvarez always uses `morph checkout`14:26
* SotK doesn't find it confusing. Checkout checks out an existing system branch, and branch creates a new system branch based on an existing system branch.14:26
* SotK always uses checkout too14:26
Zara_paulsher1ood: Ah, ok, I've tried morph help, I was just hoping there'd be a quick summary somewhere of 'ways this differs from git'; it would save me having to work it out (and possibly making mistakes!)14:26
* persia has never managed to find a useful definition for "system branch"14:26
paulsher1oodSotK: but it doesn't work - for example if you morph checkout b:b/d master14:26
paulsher1ood(all kinds of 'fun' may ensue)14:26
richard_mawif you just use checkout to build you're ok, it's the combination of checkout, b:b/d master and edit that broke things14:27
SotKpaulsher1ood: the things which cause fun there are things like `morph edit` though I think?14:27
richard_mawAIUI we were deprecating morph edit14:27
SotKIf I want to do edits, I'll do them in a new branch, not on master14:27
persiaWe had a ML thread about removing morph edit a few months ago, with the result consensus that nobody liked it and we should remove it, but nobody got around to it yet.14:28
persiaI'm not sure that counts as formal "deprecation"14:28
richard_mawwe're all aware of the need to rationalise interfaces at this point, so I'm going to keep out of this conversation and get on with this live atomic update stuff14:28
SotKIf I want to build from master, I'd like to do morph checkout b:b/d master rather than morph branch b:b/d baserock/adamcoldrick/foo14:28
persiarichard_maw: Good choice :)14:29
persiaSotK: I'd like to do `git clone definitions.git`14:29
SotKpersia: that would be wonderful, but for as long as workspaces continue to exist I like there being both checkout and branch :)14:30
Zara_thanks, persia, that was helpful :)14:31
persiaOK.  I find the existence of both confusing, and prone to causing issues if I'm not careful, so I'll continue to ignore checkout, and we can both be happy :)14:31
petefothZara_: there will be a quiz later :)14:32
*** ssam2 [~ssam2@82-70-136-246.dsl.in-addr.zen.co.uk] has quit [Ping timeout: 264 seconds]14:33
*** mdunford [~marcdunfo@82-70-136-246.dsl.in-addr.zen.co.uk] has quit [Ping timeout: 250 seconds]14:33
Zara_petefoth: as long as it doesn't involve a practical :D14:36
*** ssam2 [~ssam2@82-70-136-246.dsl.in-addr.zen.co.uk] has joined #baserock14:44
Mode #baserock +v ssam2 by ChanServ14:44
*** mdunford [~marcdunfo@82-70-136-246.dsl.in-addr.zen.co.uk] has joined #baserock14:48
*** zoli_ [~zoli_@linaro/zoli] has joined #baserock15:01
jmacsI'm getting an error while deploying: ERROR: Command failed: cp -a /src/tmp/deployments/tmpFIHbjx/tmprFW9tM/. /src/tmp/deployments/tmpFIHbjx/tmpO6khpM/tmptrecS1/systems/factory/orig/.15:12
jmacsI have 46G free on /src15:12
jmacsThen it just says "cp: write error: No space left on device15:12
jmacshundreds of times.15:12
richard_mawif you're making a disk image, then check that the DISK_SIZE in the cluster morphology is large enough15:13
petefothcurrently the only allowed valyes for BOOTLOADER_INSTALL seem to be 'extlinux' and 'none'. I guess we do ti this way so that we could install different bootloaders later on? Or are other values allowed already 15:13
jmacsThat could be it15:13
ssam2i'm taking testgerrit down for a little while to give it more disk space and more computrons15:16
pedroalvarezthat's why it's called testgerrit  and not gerrit :)15:17
persiaI rather like this use of domain names to indicate system stability.  Saves writing "This is a test system" everywhere.15:19
richard_mawI suppose it would be too silly to change its name to downgerrit.baserock.org while it's down for maintainance…15:20
persiaThe problem is that the propagation time might exceed the downtime.15:24
petefothany thoughts on http://paste.baserock.org/bubajuceki as documentation for the 'generic' write extension parameters? (DTB_PATH, BOOTLOADER_INSTALL, BOOTLOADER_CONFIG_FORMAT and  KERNEL_ARGS15:39
persiaIt might be worth noting that DTB_PATH is only interesting for systems that require DTB.15:43
persiaFor example, most x86 systems don't15:43
persiaIn BOOTLOADER_INSTALL, for "extlinux", perhaps "the extlinux bootloader will be installed", and for "none", perhaps "no bootloader will be installed".  This should make it easier to extend in the future, in line with the code architecture.15:45
persiaDo we know the interaction of "BOOTLOADER_INSTALL=none" and BOOTLOADER_CONFIG_FORMAT?  I suspect the latter is only important if the former is set, but don't know for sure.15:46
persiaKERNEL_ARGS should be optional, and explicitly indicate a default.15:46
richard_mawit's stuff to append, so the default is ''15:57
persiaappend to what?15:57
*** zoli_ [~zoli_@linaro/zoli] has quit [Remote host closed the connection]15:58
richard_mawKERNEL_ARGS is kernel command-line arguments to append to the default set which includes the btrfs subvolume and root disk15:58
persiaOr rather, the documentation should probably indicate that it is to be appended, that it is optional, and what would be there if it were empty (and how to change that).15:58
*** zoli_ [~zoli_@0x5e91887a.adsl.cybercity.dk] has joined #baserock15:58
*** zoli_ [~zoli_@0x5e91887a.adsl.cybercity.dk] has quit [Changing host]15:58
*** zoli_ [~zoli_@linaro/zoli] has joined #baserock15:58
persiapetefoth: ^^15:59
*** zoli_ [~zoli_@linaro/zoli] has quit [Remote host closed the connection]15:59
*** zoli_ [~zoli_@linaro/zoli] has joined #baserock15:59
ssam2testgerrit is back16:26
*** edmunds [~edmunds@82-71-3-239.dsl.in-addr.zen.co.uk] has joined #baserock16:45
rjekhi edmunds 16:49
edmundsHey rjek16:49
edmundsHow's tricks :-)16:49
edmundsWho is the expert around here with diskless ceph16:49
*** tiagogomes [~tiagogome@213.15.255.100] has quit [Ping timeout: 264 seconds]16:54
jmacsedmunds: Me and pdar have been working on similar systems, but they use netboot to install nodes which have local disks attached16:55
edmundsSo I want to do netboot and have them use CEPH from other machines to act as storage for them and S316:56
jmacsSo your diskless machines would be acting as a ceph front end to existing storage?16:57
edmundsYes..16:58
edmundsand then running hypversors for running VMs on16:58
jmacsThat's pretty different from what I've done tbh. All the systems I've set up configure ceph with local storage.16:59
edmundsYep... I know.. but it's quite like VMWare where the ESXi hosts are diskless and the VMDKs are on NFS or VMFS got at via iSCSI17:00
edmundsthen you can grow the hypervisior pool on demand without any issues.17:00
edmundsalso by having the OS remote you can update by moving the autofs configuration on demand or the DHCP bf and rootfs parameters17:01
KinnisonI fear you're way beyond anything we've done so far with Baserock17:02
jmacsI agree17:02
edmundsOh I'm sure currently so... but I'm trying to begin by at least showing that I can get to stateless configuration for hypervisor nodes17:03
edmundsand given we can netboot ceph.. it would be a start17:03
* persia is confused17:03
persiaIsn't this just a matter of 1) making sure you have the right PXE services enabled, and 2) making sure the systems you deploy happen to be configured diskless?17:04
* persia isn't sure how it relates to Baserock, nor why Baserock can't do that.17:05
petefoth1persia: richard_maw: thanks for the comments17:05
persiapetefoth: Thanks for the docs :)17:05
edmundsSo I believe it is as simple as building say two CEPH servers and configuring DHCP/PXE and HA to move the DHCP/PXE between the CEPH servers and then booting a linux kernel with CEPH support17:06
edmundsbut the default RedHAT or Ubuntu builds don't have this support in or the fixes to dracut to support ceph17:06
persiaSo the actual problem is just booting from a Ceph volume?17:07
jmacsOh, you want to boot the OS from a ceph volume as well17:07
edmundsi want to boot an OS from a CEPH volume17:07
*** mauricemoss_ [~simonh@82-70-136-246.dsl.in-addr.zen.co.uk] has quit [Read error: Connection reset by peer]17:09
*** mauricemoss_ [~simonh@82-70-136-246.dsl.in-addr.zen.co.uk] has joined #baserock17:09
persiaThat sounds like more a linux/Ceph issue than something we can help with directly.17:12
persiaOnce that is sorted, you should be able to use the new PXE write extension to deploy that class of system to your hardware.17:13
edmundsWell you did the diskess dracut changes and ceph boot stuff I believe didn't you17:13
persia(assuming you don't already have the servers in place).17:13
*** wdutch [~william@82-70-136-246.dsl.in-addr.zen.co.uk] has quit [Ping timeout: 250 seconds]17:13
* persia certainly didn't :)17:14
edmundsI think mdunford did :-)17:14
richard_mawI don't believe we did any dracut work. We did some cephroot stuff before, but that was for putting bits in the kernel, not an initramfs17:14
edmundsI want the OS repository on CEPH to be RO.. like to do with ReadOnly (RO) NFS boot17:15
richard_mawI think you mean doffm rather than mdunford17:15
edmundsI think you're right I do mee doffm17:15
richard_mawI don't know what happened to putting cephroot in the kernel. The patch submission thread is https://lkml.org/lkml/2013/11/20/630, but I don't have time to look at it right now17:18
richard_mawthough if doffm were to come in and answer some questions that would be best17:18
KinnisonI think it was decided there are better ways to do it17:18
Kinnisoninvolving initramfsen17:18
edmundsI thought it had been accepted and advanced..17:22
edmundsI assumed that the dracut changes supported initramfs necessities17:22
richard_mawI don't recall us ever doing anything with dracut17:25
persiadracut not being lorried would tend to support that recollection17:27
edmundsArhh ok17:28
ssam2ANNOUNCEMENT: git.baserock.org will be down for maintenance at 18:30 UTC17:42
ssam2for hopefully no more than 30 minutes17:42
persiaThat includes the reboot by the provider and the maintenance?17:50
*** Krin [~mikesmith@82-70-136-246.dsl.in-addr.zen.co.uk] has quit [Remote host closed the connection]17:56
*** edmunds [~edmunds@82-71-3-239.dsl.in-addr.zen.co.uk] has quit [Read error: Connection reset by peer]17:57
*** bashrc [~motters@82-70-136-246.dsl.in-addr.zen.co.uk] has quit [Quit: Lost terminal]18:01
*** zoli_ [~zoli_@linaro/zoli] has quit [Remote host closed the connection]18:15
ssam2persia: yes18:18
ssam2git.baserock.org will be down in 1 minute18:29
pedroalvareznote: cache .baserock.org has been mirroring git.baserock.org, so can be used meanwhile git.baserock.org is down18:30
*** mdunford [~marcdunfo@82-70-136-246.dsl.in-addr.zen.co.uk] has quit [Ping timeout: 250 seconds]18:38
*** locallycompact [~lc@82-70-136-246.dsl.in-addr.zen.co.uk] has quit [Ping timeout: 244 seconds]18:39
ssam2git.baserock.org is back and shiny and new19:01
*** tiagogomes [~tiagogome@host-92-25-187-78.as13285.net] has joined #baserock19:01
*** tiagogomes [~tiagogome@host-92-25-187-78.as13285.net] has quit [Client Quit]19:01
ssam2git server, gitweb, lorry and gitano all seem to be running19:01
ssam2good night all!19:03
*** ssam2 [~ssam2@82-70-136-246.dsl.in-addr.zen.co.uk] has quit [Quit: Leaving]19:03
persiaThanks ssam2!19:05
*** SotK [~adamcoldr@access.ducie-dc1.codethink.co.uk] has quit [Remote host closed the connection]19:16
*** benbrown_ [~benbrown@access.ducie-dc1.codethink.co.uk] has quit [Read error: Connection reset by peer]19:16
*** jmacs [~jimmacart@access.ducie-dc1.codethink.co.uk] has quit [Remote host closed the connection]19:16
*** kejiahu [~kejiahu@access.ducie-dc1.codethink.co.uk] has quit [Read error: Connection reset by peer]19:16
*** perryl [~laurenper@access.ducie-dc1.codethink.co.uk] has quit [Read error: Connection reset by peer]19:16
*** Zara_ [~zarazaime@access.ducie-dc1.codethink.co.uk] has quit [Read error: Connection reset by peer]19:16
*** petefoth1 [~petefothe@access.ducie-dc1.codethink.co.uk] has quit [Read error: Connection reset by peer]19:16
*** DavePage [~dpage@access.ducie-dc1.codethink.co.uk] has quit [Read error: Connection reset by peer]19:16
*** pdar [~patrickda@access.ducie-dc1.codethink.co.uk] has quit [Read error: Connection reset by peer]19:16
*** mwilliams_ct [~mikewilli@access.ducie-dc1.codethink.co.uk] has quit [Read error: Connection reset by peer]19:16
*** paulsher1ood [~paulsherw@access.ducie-dc1.codethink.co.uk] has quit [Read error: Connection reset by peer]19:16
*** ridgerunner [~robjones@access.ducie-dc1.codethink.co.uk] has quit [Read error: Connection reset by peer]19:16
*** genii [~quassel@ubuntu/member/genii] has joined #baserock19:19
*** mwilliams_ct [~mikewilli@access.ducie-dc1.codethink.co.uk] has joined #baserock19:35

Generated by irclog2html.py 2.14.0 by Marius Gedminas - find it at mg.pov.lt!