[00:36] I've got two ubuntu servers talking via cifs but i'm having an issue where some services that I have running are not having the right permissions when trying to write back to the cifs share [00:37] storrgie_, CIFS is not a POSIX filesystem [00:37] storrgie_, you should be using NFS [00:38] escott, you're right... is there a way to force permissions to work over samba though? probably in version 4 [00:38] escott, I am not running kerberos... is nfs just plain scary to run if you dont have an auth service? [00:39] storrgie_, maybe http://www.samba.org/samba/CIFS_POSIX_extensions.html [00:39] escott, If you're good with networking you might be able to help me. the systems I'm talking about are actually a host and VM [00:39] so maybe i can define a virtual interface for them to talk through that will be protected [00:39] storrgie_, install the guest additions [00:39] im using kvm [00:40] storrgie_, http://dustymabe.com/2012/09/11/share-a-folder-between-kvm-host-and-guest/ [00:42] escott: wow. I had no idea such a thing was doable. neat. thanks. :) [00:42] yeah its the 9p thing [00:42] im scared to do that [00:42] but i probably should be doing it [00:42] im struggling because i want to use CIFS though [00:42] so i only have one service to share info with [00:42] storrgie_, i would certainly NOT do this with CIFS [00:43] storrgie_, NFS or plan9 seem to be the better options for kvm. kinda sad they dont have a direct paravirt setup [00:43] escott, I'm not sure how to lock down NFS though [00:44] storrgie_, lock down meaning? [00:45] storrgie_: you could have your /etc/exports file restrict the export to only the one guest IP.. [00:45] with nfs dont you just say any ip can come in [00:45] well any one [00:45] so here is the setup [00:45] its a KVM host that im trying to get to [00:45] could I just define a bridge interface and add that to my VMs to use [00:45] so its a private network between them [00:46] yes [00:46] though if you want your vms to talk to each other, you'll probably have some more work to do :) [00:46] they already have one bridge interface to get to the network [00:46] sarnold, do you know the convention for doing this though [00:46] my first bridge is tied to a physical interface [00:47] this bridge would be virtual [00:48] storrgie_: sadly, no; I just know enough to know it's possible to create a new bridge between the host and one specific guest that the other guests can't get to [00:49] storrgie_: I _think_ the virsh net-* family of functions is the starting point.. [00:49] nah i dont want to use virsh [00:49] I wnt to define a bridge in interfaces [00:51] storrgie_: ah, not using libvirt for the kvms then, okay... brctl is probably the place to start :) [00:52] yeah, im looking at virsh though also [00:52] hrmm [00:55] sarnold, do you know much about creating bridges? [00:55] I would like to define it in a file [00:55] rather than run brctl [00:55] there is a page here that describes that I want... I think http://www.linux-kvm.org/page/Networking [00:56] I just dont know how to convert that to an addition to my /etc/networking/interfaces file [00:56] storrgie_: sadly, no, this is one of my weakest areas. [00:58] pewp [00:58] well here ill post once more and see if anyone bites: [00:58] I'm trying to define a private bridge for my KVM vms to talk to my host. I want to define it in my /etc/networking/interfaces file. I found this document here: http://www.linux-kvm.org/page/Networking that I think talks about what I want to do but I don't understand the brctl commands enough to decompose those into additions to my /etc/networking/interfaces config file [01:56] Hi everyone [01:57] How would one go about routing bittorent traffic through an openvpn connection but keep http, dns ect through my isps connection? [01:58] Would this be done by changing some routing on my ubuntu server? [02:03] Amacidia, routing is at the ip level, you want a port specific rule. you could do that with iptables === Ursinha is now known as Ursinha-afk === Ursinha-afk is now known as Ursinha === cpg|away is now known as cpg === acidflash_ is now known as acidflash [04:53] hi [04:56] holas [05:25] anyone know how to activate the UI on mediatomb using a headless ubuntu server? [05:25] I'm having trouble finding appropriate documentation [05:26] im logging into the server via ssh [05:27] chovynz: isn't it just a web interface? [05:27] runs on port 49152 by default [05:28] I guess. this is the web message i get "MediaTomb UI is disabled. Check your configuration." , I guess my question is more to do with mediatomb than US, but [05:29] most of the documentation i can find on mediatomb talks about doing things via ui [05:29] chovynz: probably need to edit your /etc/mediatomb/config.xml [05:29] [05:29] ^^ mine has that [05:29] pleia2: Error: "^" is not a valid command. [05:30] oh ok. thanks === Ursinha is now known as Ursinha-afk [05:31] sure, it's not the easiest thing to configure (particularly when it comes to codecs and things) but there are a lot of example config.xml files out there to get tips from [05:33] *cough*ex-windows-user-here-who-still-hasn't-got-it-completly-out-of-his-system-of-getting-things-done-for-him*cough* [05:33] :) [05:33] thanks pleia2, i didn't think of that [05:33] sure, good luck [05:34] uh, another silly question... could the reason the ui is disabled, is that i haven't yet hooked it up to a database? [05:34] could be [05:35] it's been a long time since I configured mine, I don't really remember the steps involved [05:35] ahaha.. me too [05:35] I'm using sqlite though === Ursinha-afk is now known as Ursinha [05:51] New bug: #1091523 in qemu-kvm (main) "Testing Raring in KVM displays visual artifacts (affects installer world map, login screen, wallpaper)" [Undecided,Confirmed] https://launchpad.net/bugs/1091523 [06:19] guys is there anyway to call vmbuilder and output the results onto an nfs share? [06:19] i've done -d and it never seems to complete.. [06:19] (yes I have r/w) [06:59] wow not everyone at once... [07:15] can i set 2 listening ports for sshd? === _ruben_ is now known as _ruben [07:34] Hi , Apache in my server doesn't load index.php automatically ! I tried DirectoryIndex index.php index.html index.htm but that was helpless [07:35] can u help me please ? [07:35] do you have php installed [07:35] yes , it is work when i write index.php [07:36] ex : www.ex.com dosen't work but www.ex.com/index.php it fine [07:39] perhaps you can't override DirectoryIndex with a local .htaccess [07:40] i tried to do that in httpd.conf AND sites-available/default [07:44] I want from my server to execute a command on it's start, Could you please let me how to this? [07:45] *let me know [07:49] Syria http://www.cyberciti.biz/tips/linux-how-to-run-a-command-when-boots-up.html === ibiris|afk is now known as ibiris [07:51] hath001: Thank you I will read it. [07:55] hath001: The script name.sh should contain the commands I want them to be executed at the startup only? [07:56] hath001: Sorry I see it now. [08:20] I have a server that I also want to use as a desktop sometimes, but I don't want _everything_ that comes with ubuntu-desktop meta package. Is there a way of installing ubuntu-desktop where you can pick and choose the components? [08:27] oracle_, pick and choose the individual packages you want to install? if you install the metapackage it'll install everything, if you hand-pick what you want from the list of what the metapackage installs, you'll get the customized list of stuff. [08:28] i don't remember how extensive the list of stuff is though [08:30] Are some packages reverse-dependant on ubuntu-desktop? So that even if I install OOo separately, it will drag in ubuntu-desktop as a dependency? [08:30] sorry, OOo is now LibreOffice i suppose. [08:42] jamespage: tracking https://lists.launchpad.net/openstack/msg19520.html ? [08:42] mornin' btw [08:44] morning chaps, any recommendations for the best way to set a 'one time reboot' for a server. E.g. Security updates applied automatically, system needs a reboot and I want to set it to happen 00:00 tonight? [08:49] feisar: the shutdown command can take a timestamp for when to do it [08:52] balboah: ah, that sounds good, I could write a little script that asked for the time and then issued the command [08:53] feisar: sure. You can also use the "at" program do run a certain script one time at a certain time [08:53] would I have to issue the command from a Screen session (or similar) or would the command still execute if I closed the terminal? [08:53] it will be scheduled and executed even if you close your terminal [08:55] Daviey, morning [08:55] I had seen that yes [08:55] nothing new tho [08:55] we are using most of that stuff [08:57] balboah: thanks : ) [08:58] Daviey, did you do the NEW ack's for the latest ceph packages in raring? === schmidtm_ is now known as schmidtm [09:08] jamespage: no, when i signed off they were not al built === yofel_ is now known as yofel [09:08] jamespage: but it is done now [09:09] jamespage: and it's now in the release pocket [09:09] Daviey, it is - but a few bits and pieces have crept into main [09:09] I'd like them pushed back to universe [09:10] jamespage: hmm, it's not showing as a candidate for demotion. [09:10] http://people.canonical.com/~ubuntu-archive/component-mismatches.txt === mcclurmc_away is now known as mcclurmc [09:15] Daviey, I don't think it will [09:16] Daviey: ceph-fuse, libcephfs-jni and libcephfs-jni-dbg should be pushed to universe [09:16] jamespage: then something is keeping it in main :) [09:17] hmm [09:18] Daviey, I don't think so - nothing rdepends on ceph-fuse or libcephfs-jni [09:22] hmm [09:25] jamespage: You indeed seem to be correct, as usual. [09:26] jamespage: lets see if anything demands it back, http://pb.daviey.com/iMRD/ [09:27] Daviey, ceph-fuse also has a -dbg - that will demand it back [09:29] `less *.log` then :n and :p to move back and forth through the files. Can I improve that down to a single keystroke for next / previous file? [09:29] With more maybe? Or some other tool? [09:29] I guess I could use vim and map a keystroke to next/prev file, but I'm looking for something that'll work out of the box, on almost any platform... :-) [09:30] New bug: #1091577 in samba (main) "unable to ping hosts by windows hostname" [Undecided,New] https://launchpad.net/bugs/1091577 [09:31] jamespage: done [09:32] Daviey, muchly [09:32] ta [09:32] sorry - that came out in the wrong order [09:33] jamespage: problem no ;) [10:26] New bug: #1025244 in qemu-kvm "qcow2 image increasing disk size above the virtual limit" [High,Triaged] https://launchpad.net/bugs/1025244 [10:28] New bug: #1091602 in bind9 "Please add resolvconf hook script to generate dynamic forwarders list" [Undecided,New] https://launchpad.net/bugs/1091602 [11:45] New bug: #1091624 in net-snmp (main) "package snmpd 5.4.3~dfsg-2.5ubuntu1 failed to install/upgrade: subprocess installed post-installation script returned error exit status 127" [Undecided,New] https://launchpad.net/bugs/1091624 [13:09] zul: around [13:14] having this problem running ceilometer on devstack: http://paste.ubuntu.com/1447493/ [13:14] any idea? [13:34] i am trying to configure iscsi server on ubuntu as a target (as a server )(please correct me if i am wrong )with esxi server [13:38] samba35: should work [13:39] but when i run tgt-admin -s it print noting [13:39] do you have any idea what configration changes i have to make [13:39] to be honest, I've never setup iscsi target on linux - I have this test box, though - lemme see :) [13:40] ok [13:40] no problem ,but please correct me [13:41] well, tgt-admin -s should show what's setup [13:42] initiator is know as a client and target is know as a server right [13:42] yes [13:42] thanks [13:43] is there any gui for iscsi i ubunut ? [13:44] probably not [13:44] ok [13:45] are you following a howto? this one looks decent http://askubuntu.com/questions/63621/iscsi-trgt-missing-because-cant-build-kernel-module [13:46] now i will === mcclurmc is now known as mcclurmc_away === cpg is now known as cpg|away === matsubara is now known as matsubara-lunch === mcclurmc_away is now known as mcclurmc [14:56] hey all, my /boot filled up and i've got 800 megs free and am getting dpkg: error processing /var/cache/apt/archives/linux-headers-3.2.0-35-generic_3.2.0-35.55_amd64.deb (--unpack): [14:56] unable to create `/usr/src/linux-headers-3.2.0-35-generic/include/config/acpi/cmpc.h.dpkg-new' (while processing `./usr/src/linux-headers-3.2.0-35-generic/include/config/acpi/cmpc.h'): No space left on device any ideas? [14:56] i used dpkg to purge some stuff and now boot is at 50% and / shows its 80% with 800 megs free and it is still getting that no space left on device error === Ursinha is now known as Ursinha-afk [15:03] and i created a 100 meg file there just fine with dd [15:07] hi there somebody here experienced with webmin? === Ursinha-afk is now known as Ursinha [15:29] when i try to add service "chkconfig anyservice on" i am getting /sbin/insserv: No such file or directory [15:30] samba35: chkconfig isn't used on Ubuntu. [15:30] !webmin [15:30] webmin is no longer supported in Debian and Ubuntu. It is not compatible with the way that Ubuntu packages handle configuration files, and is likely to cause unexpected issues with your system. [15:30] then how do i start service at startup [15:31] samba35: Which service? [15:31] samba35: it's usually started by upstart on startup if it's installed. which service is this? [15:31] say vsftpd tgt or any [15:31] samba35: well, not 'any' :) [15:32] when i run chkconfig --list it show services [15:32] samba35: don't use chkconfig [15:32] samba35: Ubuntu uses upstart/init.d. [15:32] to start tgt, run 'start tgt' [15:32] tgt is in upstart [15:32] ok [15:32] so is vsftpd [15:33] ok [15:33] thanks [15:33] did you manage to make tgt work? [15:33] no [15:33] * RoyK has been out for a while [15:33] will try [15:33] ok === matsubara-lunch is now known as matsubara [15:35] samba35: http://www.cyberciti.biz/tips/howto-setup-linux-iscsi-target-sanwith-tgt.html [15:36] samba35: the other I posted was for another iscsi target implementation - an older one I think [15:36] * RoyK didn't know there were two [15:36] ok let me try [15:36] thanks [15:45] i think i fail to understand "iqn" [15:46] that's just the iscsi initiator name prefix [15:46] iSCSI Qualified Name [15:48] but how to use full name [15:48] do i have to use host or domain ? [15:49] it's usually something like this iqn.1993-08.org.debian:01:a44148977e36 - meaning iqn.YYYY-MM.tld.domain:something [15:49] it should be autogenerated when you install open-iscsi [15:50] how do i generate it ? [15:50] apt-get install open-iscsi open-iscsi-utils [15:50] that's for the initiator, though [15:50] no need for an initiator name on the target side [15:50] ok [15:51] what OS does the initiator run? [15:51] or 'on what what OS does the initiator run?' [15:52] * RoyK just created a 200GB iscsi volume and attached to it from localhost :) [15:52] i am trying windows xp and vmware esxi [15:52] ok, both should work [15:52] i have to use server iqn at client ? [15:53] each client has its own iqn [15:53] on the server, you can authorize a given iqn for a given target, you can authorize an ip address, or setup authentication [15:53] s/server/target/ [15:53] confued :) [15:53] well, it'll sink in :) [15:54] took me some time to get used to how iscsi works [15:54] that is [15:54] all targets have an iqn, all clients have one as well [15:54] s/clients/initiators [15:55] that is, all endpoints in an iscsi setup, has an iqn [15:55] RoyK, do you use vmware esxi [15:55] at work, yes, on iscsi [15:55] I guess it should autogenerate its iqn as well [15:55] for testing, just allow all clients: tgtadm --lld iscsi --op bind --mode target --tid 1 -I ALL [15:55] esxi has a target or client ? for version of esxi [15:55] utlemming, thought i'd point you to 'mount-callback-umount', which mounts a disk or partition image in anything qemu-nbd supports and hten calls a command and unmounts. [15:55] its at https://code.launchpad.net/~smoser/+junk/backdoor-image [15:56] i'm using it for things like this: http://paste.ubuntu.com/1447788/ [15:56] esxi has an initiator (that is, the client). esxi shouldn't need a target (since it doesn't serve storage) [15:56] which mounts an image, installs a package or 2 and then publishes it. [15:57] unfortunately, grub and kernel cause issues with a simple 'mount-callback-umount chroot MOUNTPOINT sh -c "apt-get update && apt-get dist-upgrade"' [15:57] ok [15:59] a 'target' is a 'server' meaning it serves storage, an 'initiator' is the one using the storage. this is the same as the old SCSI controller is a 'target' where the OS (or drivers) are 'initiators' [16:01] ok [16:01] i am trying ubuntu as a target to esxi [16:02] if you want to do some testing first, apt-get install open-iscsi open-iscsi-utils and use that to check if you can see the target [16:03] yes i have installed this packages [16:03] need to fix iqn [16:03] such as http://paste.ubuntu.com/1447815/ [16:04] samba35: what does /etc/iscsi/initiatorname.iscsi say? [16:05] InitiatorName=iqn.1993-08.org.debian:01:51e9ab41fb7e [16:05] looks like a valid IQN to me [16:06] what is my iqn then [16:07] that's the initiator name [16:07] the targets will have an iqn each [16:07] ok [16:08] as in my example above [16:08] my iscsi test target is named iqn.2012-12.net.karlsbakk:iscsitest [16:08] send pm [16:08] pm answered [16:09] thanks [16:09] did you genrate this iqn ? [16:10] basically, use tgt to setup a target, that is, with a number and name, after that, assign a partition or lvm logical volume, any device or file, really, to that target, and set ACLs on that [16:10] this is in your /etc/iscsi/initinatoname.iscsi ? [16:10] that iqn was generated in my head, "iqn.YYYY-MM.net.karlsbakk:targetname [16:11] no, the initiator and target are different endpoints and must therefore have different iqns. I use the one (like yours) autogenerated [16:11] how do i detect my iqn for target [16:12] have you connected with iscsiadm -m discovery -t st -p localhost ? [16:13] no [16:13] first, you need to do discovery [16:13] then you can login to a target [16:14] ok [16:14] so, pastebin tgt-admin -s first [16:14] its blank [16:14] then you need to setup a target [16:14] as described on http://www.cyberciti.biz/tips/howto-setup-linux-iscsi-target-sanwith-tgt.html [16:14] please read that through first [16:14] ok [16:31] utlemming, uploading new ec2-api-tools to raring [16:33] samba35: any more luck? [16:33] moveing forward slowly [16:34] the proble is my ubuntu is on vmware and behind a firewall [16:34] they both are on vm and 4 gb ram and 3 guest [16:34] smoser: ack. I wonder if caribou wants to drive SRU's for that :) [16:35] its not worth sru imo. [16:35] samba35: erm - you're using a VM to serve storage to esxi? [16:35] smoser: yeah, I know.... :( . But worth a PPA build [16:35] sru's are a pain in the ass, and slittle things like this that are easily got from a ppa only slwo up the queue for important things. [16:35] for testing 200 mb only [16:35] ok [16:36] should work for testing, I guess... [16:36] utlemming: ??? [16:36] utlemming: is that a xmas gift ? ;-) [16:37] caribou: I'm a giving sort of guy :) [16:37] utlemming: I'd say let's wait until someone asks for it [16:37] caribou: just giving you a hard time...you seem to do all the SRU's lately that I watch [16:38] well, this is kind of my dayjob (at least the one I'm aiming for) [16:38] utlemming: got another one coming on lvm2 [16:41] do i need to worry about my nic whether it support iscsi or not ? [16:42] trying to add iscsi with esxi now there is no movement from esxi server is busy [16:42] (mouse busy sign ) [16:43] iscsi runs over tcp [16:43] so any nic will do [16:43] ok [16:44] utlemming, just tested that snapshot copy works. [16:44] it returns immediately [16:44] is there any way to test from ubuntu [16:45] then you have to wait until it gets completed before you can register. [16:45] samba35: for testing, that is - for a good setup with large amounts of storage, you'd want checksumming on the NIC [16:45] samba35: yes, see above - iscsiadm etc [16:45] smoser: that's what I suspected....but this would substaintly cut down on the time to register [16:45] maybe. [16:45] probably [16:45] smoser: when the current fire ends, I'll dig into using migrate-image and migrate-snap [16:46] but our wget and populate is pretty fast. [16:46] smoser: I wonder about ap-southeast-{1,2} transit [16:46] utlemming, i'd suggest wrapping it in a "snapshot-and-register" [16:47] that does what we actually *want* [16:47] ok [16:47] which can then be replaced with their "Amazon Machine Image (AMI) Copy" that they've suggested is coming. [16:50] utlemming, also, if you wrap it, make it correctly copy description (by default). [16:50] as they do not do that right now [16:53] samba35: http://paste.ubuntu.com/1447945/ [16:54] samba35: I first created a 200GB lv, named /dev/raid/iscsitest [16:54] vmware normally even performs better using software iscsi vs using an iscsi nic [16:55] reasons being, you can't loadbalance iscsi over multible hardware iscsi nics [16:55] vmware's multipath implementation is rather good after what I hear [16:55] it just got a huge tuneup in 5.1 :) [16:55] we're updating all hosts to 5.1 these days [16:56] larger clusters [16:56] I've just finished basically [16:56] and virtual datacenter one day, I guess [16:59] patdk-wk: have you looked at the virtual datacenter stuff in vmware? [16:59] * RoyK is obviously aware of this being rather [offtopic] [17:05] jamespage: hey james, how would a merge like this work, if its just a bare packaging branch being merge into the ubuntu branchd /w no upstream source included and no remote tarball for the version [17:05] https://code.launchpad.net/~openstack-ubuntu-testing/nova/precise-essex-proposed/+merge/140225 [17:05] adam_g, badly infortunately [17:06] samba35: did you take a look at that post? [17:06] adam_g, I worked with yolanda on getting that working better earlier today [17:06] New bug: #1091763 in lxc (universe) "lxc-start-ephemeral freezes (dup-of: 1021471)" [Undecided,New] https://launchpad.net/bugs/1091763 [17:06] it'd be rather nice if someone came up with a vCenter like interface to KVM... [17:06] adam_g: she raised new MP's based on the distro branches by importing the source packages [17:06] jamespage: ohh. what did you come up with? i've been manually generating tarballs [17:06] adam_g, https://code.launchpad.net/~yolanda.robla/+activereviews [17:07] adam_g, well - its still manual to create the tarball; but by doing a 'bzr import-dsc' into a full packaging branch you end up with something that can be proposed for merge [17:07] server is busy still i think i have to reboot [17:07] sorry [17:07] adam_g, and a sponsor can grab a test as it contains the orig.tar.gz [17:08] the reasom i am not rebooting is i am download ing some stuff [17:08] RoyK: there is the libvirt gui being ported to a win32 frontend and someone is trying to port it to ipad at the moment [17:09] RoyK: it's not a bad starting place for a "v center style front end" [17:09] jamespage: i see [17:12] ikonia: it'd be nice if virt-manager was a bit better - but then - I'm not a coder, so please don't blame me ;) [17:13] RoyK: I like it, it's a great starting place [17:14] if it gets ported better to windows and tools such as ios/android it will have a sound place to expand [17:14] adam_g, reviewing swift_redux branches now [17:14] ikonia: I use it regularly myself - no problem - it's just that once I started working with vmware, I god a bit surprised how much better it was [17:14] jamespage: cool. [17:15] RoyK: yeah, the vmware tools are a big step up, but then, so is the development budget / time etc [17:15] ikonia: I'm aware of that ;) [17:16] for small busines or even certain enterprise application it ticks the boxes just fine [17:16] RoyK: Redhat's virtualization platform has really expended on it too [17:16] these edu licenses we get cost about $2k per node - no idea how much that costs for non-edu organizations [17:16] so if RedHat feed their updates in to, so other distros such as ubuntu can benifit that would make a big jmp [17:16] RoyK: 5K-ish [17:16] *if* redhat chooses to share that code [17:17] (functionality/tin dependent) [17:17] 5k per node is rather a lot for a 10-node cluster [17:17] RoyK: sorry, that's not right, it's not per node in hte cluster [17:17] it's per vcenter [17:18] the individual nodes are much less [17:18] erm - that's just the admin tool [17:18] (or whatver Vcenter is calling itself this week) [17:18] or node [17:18] I think it's still vcenter [17:18] wht's the word for the "enterprise estate" [17:18] it's not "estate" [17:18] I can't fire up the gui to look now [17:18] "farm" but not "farm" [17:19] for 10 nodes, a single vcenter installation works well [17:19] that's not what I meant, I didn't actually mean the vcenter interface [17:19] the "estate" but I can't think of the word [17:20] what interface is that? the gui? [17:20] * RoyK is somewhat annoyed with Dell Equallogic storage [17:21] yes, it's a gui, but it's basically you pay $X per "estate" for all the functionality like vmotion, etc, and then %Y per node in that estate [17:21] but "estate" isn't the word, [17:22] but what does it mean? [17:22] it used to be the vcenter "suite" [17:23] it's the funcionality that allows for vmotion migration, on the fly storage changes, etc [17:23] that's why I was refering to it as vcenter, I didn't mean the actual gui but the enterprrise tools it provided ontop of the virtualization [17:23] it's basicaly bolt on enterprise functions to vsphere. [17:23] but I can't think of the word it's called [17:23] sure, but how many of those would you need? with 10 nodes? [17:24] 1 [17:24] you buy one and then the individual nodes are licensed as vsphere/whatever nodes [17:25] * RoyK flags #ubuntu-offtopic for vmware talk [17:32] adam_g, OK - so how do I get the ring to rebalance and actually start ? [17:35] jamespage: initially? you need a number of zones >= the number of minimum replicas configured (3 by default) [17:35] jamespage: i can extend the README if its unclear [17:35] adam_g, no - I'm being a bit dumb [17:36] who reads a README anyway [17:36] :-) [17:36] jamespage: in the case of 3 minimum replicas: if configured to do manual zone assignemnt, you'd need 3 swift-storage service units each configured as their own zone. with auto, you'd need 1 storage unit with 3 machine units [17:36] adam_g, I have three but they are all in the same swift-storage service and I'm using manual [17:39] jamespage: you have 3 what? 1 swift-storage service with 3 machine units? [17:39] adam_g, yes [17:39] in that case you only have 1 storage zone. you need to add two more swift-storage service units, configured as zone: 2 + zone: 3 [17:42] jamespage: if you dont want to follow the charm.log, you can run 'swift-ring-builder /etc/swift/account.builder' to where nodes are being assigned === VD is now known as Guest44874 [17:46] adam_g, OK - think I have it right now [17:46] adam_g, the interfaces need better typing [17:47] jamespage: yeah, i was gonna wait for a review to get feedback on that. i was trying to come up with better interfaces but was having a hard time. partly because i dont fully grok juju interfaces [17:50]