Perl 6 - the future is here, just unevenly distributed

IRC log for #fuel, 2015-03-25

| Channels | #fuel index | Today | | Search | Google Search | Plain-Text | summary

All times shown according to UTC.

Time Nick Message
00:28 rmoe joined #fuel
00:51 emagana joined #fuel
00:53 emagana_ joined #fuel
01:09 Longgeek joined #fuel
01:35 mattgriffin joined #fuel
01:51 xarses joined #fuel
01:59 mattgriffin joined #fuel
02:54 kevinbenton joined #fuel
02:59 Longgeek joined #fuel
03:38 claflico joined #fuel
03:52 ilbot3 joined #fuel
03:52 Topic for #fuel is now Fuel 5.1.1 (Icehouse) and Fuel 6.0 (Juno) https://software.mirantis.com | Fuel for Openstack: https://wiki.openstack.org/wiki/Fuel | Paste here http://paste.openstack.org/ | IRC logs http://irclog.perlgeek.de/fuel/
04:00 Longgeek joined #fuel
04:45 zerda joined #fuel
04:55 claflico joined #fuel
05:10 rzerda joined #fuel
05:10 omolchanov joined #fuel
05:16 Longgeek joined #fuel
05:49 emagana joined #fuel
06:07 rainbowzerda joined #fuel
06:50 bazilza joined #fuel
07:03 dklepikov joined #fuel
07:12 Longgeek joined #fuel
07:41 bazilza joined #fuel
07:53 hyperbaba joined #fuel
08:01 bazilza this bug references fix but none of the urls are working https://bugs.launchpad.net/mos/+bug/1371723
08:01 bazilza how do i get the new package?
08:29 e0ne joined #fuel
08:43 HeOS joined #fuel
08:49 Miouge_ joined #fuel
08:57 adanin joined #fuel
09:07 corepb joined #fuel
09:07 Longgeek joined #fuel
09:07 corepb joined #fuel
09:17 saibarspeis joined #fuel
09:32 bazilza joined #fuel
09:43 alecv joined #fuel
09:43 neophy joined #fuel
09:54 azemlyanov joined #fuel
10:01 corepb__ joined #fuel
10:07 e0ne joined #fuel
10:09 Miouge joined #fuel
10:15 corepb joined #fuel
10:21 Longgeek joined #fuel
10:38 neophy_ joined #fuel
11:02 BobBall Is there a way to force the overlap of some networks?  I've got a /24 that I want to use for PXE, floating and public IP addresses.  I can easily set the ranges for each to not overlap, but the check-networks test fails explaining the various overlaps
11:07 e0ne joined #fuel
11:08 f1gjam2 HEy Bob
11:08 BobBall Howdy :)
11:09 f1gjam2 How many interfaces are you using?
11:09 f1gjam2 Since the PXE network is untag - are you using VLAN's
11:09 BobBall 1 - but I have plenty of VLANs.  The issue is the routable-nature of the IP ranges.  I have one routable range and in order to pass the health checks at least floating + public need to be routable
11:10 BobBall Actually I can have as many interfaces as I need
11:10 f1gjam yes thats correct
11:10 f1gjam ok - so the way I did this was
11:10 BobBall (Running virtualised :) )
11:10 f1gjam I used 192.168.10.0/24 - for public and floating
11:10 f1gjam then in the environment setup
11:10 BobBall Are you allowed an overlap between public + floating?
11:11 f1gjam you dont overlap - you have to state what the range is
11:11 f1gjam so i do the following
11:11 f1gjam I say 192.168.10.127-253 - floating - with 192.168.10.254 as GW
11:11 BobBall Ah; understood.  So perhaps if I put PXE on a non-routable VLAN and just have public + floating sharing that range you think it'll work?
11:11 f1gjam I say 192.168.10.127-253 - Publick- with 192.168.10.254 as GW
11:11 f1gjam I say 192.168.10.127-253 - Public- with 192.168.10.254 as GW
11:11 f1gjam and then
11:12 f1gjam 192.168.10.1-128 - floating
11:12 f1gjam you still define the network as 192.168.10.0/24
11:12 f1gjam does that make sense?
11:13 BobBall *tests*  With _another_ reinstall because you can't change the master node IP address post-install! :)
11:13 f1gjam Nope
11:13 f1gjam if you want i can send you a screen shot link
11:13 f1gjam of the setup?
11:13 f1gjam ive done in regards to network?
11:13 BobBall Might be useful, just for reference, please.
11:13 f1gjam one sec
11:14 f1gjam https://dl.dropboxusercontent.com/u/833212/Screenshot_032515_111427_AM.jpg
11:15 f1gjam kaliya
11:16 f1gjam kaliya - I spoke to support they said you HAVE to use the fuel master node as a jump host to get to the controller and compute node. I am a bit disappointed as another support guys said you dont have to, as I belive you did also. So i think there is some confusion if you have to or not?
11:19 bazilza f1gjam, if you assign public ips to nodes you don't have to. there is an option to do so in fuel settings when you setup the environment
11:20 kaliya f1gjam: if you're SSHing from a host on the same L2segment or with access to nodes public IPs, why do you want to jump through the Master?
11:26 f1gjam what I am doing is trying to access the nodes from 192.168.1.x to the admin/pxe network 192.168.20.x
11:26 f1gjam via the gateway 192.168.20.x
11:27 f1gjam bazilza I havent assigned any public addresses to the physical hosts (other then what the fuel installer does for the controller node)
11:28 f1gjam I wanted to treat the admin/pxe network like any other network, however because all the nodes on openstack point to the fuel master for GW this doesnt seem possible
11:29 f1gjam so in my case the only node i can reach via the LAN is the fuel master
11:29 f1gjam admin/pxe network should be routed
11:36 f1gjam i am positive this is a bug
11:39 stamak joined #fuel
11:40 ricolin joined #fuel
11:45 bazilza joined #fuel
11:49 corepb joined #fuel
12:18 mattgriffin joined #fuel
12:18 mattgriffin joined #fuel
12:19 mattgriffin joined #fuel
12:21 Longgeek joined #fuel
12:33 BobBall f1gjam: How do your the network settings work?  It seems that 192.168.10 is on VLAN 10 but the floating IP ranges don't let you specify a VLAN?
12:34 f1gjam you can specify the vlan
12:34 f1gjam in my case its vlan 10
12:34 f1gjam oh
12:34 f1gjam the floating you dont need to
12:34 BobBall Under the floating IP ranges you don't get to specify a VLAN
12:34 BobBall does it inherit the VLAN from public?
12:34 f1gjam thats correct you dont need to
12:34 f1gjam yes that correct
12:34 f1gjam floating is a slice of public
12:34 f1gjam you just need to make sure the ranges are not overlapping
12:35 BobBall Ah! That would _SO_ be more obvious if you defined floating in the same group as public...
12:37 corepb joined #fuel
12:39 f1gjam ;)
12:40 f1gjam it could be a lot clearer i agree
12:40 BobBall Can nodes have multiple physical devices?
12:40 f1gjam network devices?
12:40 BobBall yes
12:40 f1gjam ues but you need to make sure you have the vlans setup correctly
12:40 f1gjam i only use 3 interfaces
12:41 f1gjam its the best openstack installer by far, just needs a little tweaking
12:42 BobBall Ah - I see how it works
12:42 BobBall OK - I think I can make this work.  Maybe.
12:46 BobBall Can I get fuel to remove nodes so they can be rediscovered?  I've added an extra NIC to my nodes but Fuel still only seens 1 interface
12:47 BobBall oh - timing delay.  Seems it's found the interface now, sorry.
12:53 dilyin joined #fuel
13:03 f1gjam yes
13:03 f1gjam you can :)
13:03 f1gjam you can keep deleting and adding nodes it should rediscover, however it reboots the nodes and re-pxe's them once the boot is complete is shows up in fuel
13:04 BobBall yah... am deploying now.  After I found the network configuration section it all seems to be working nicely.  I understand now and should be able to coax it into my environment
13:09 kaliya joined #fuel
13:22 Longgeek joined #fuel
13:30 angelberrios hi!
13:32 angelberrios There is any way to reset the changes column in the Postgres DB?
13:32 angelberrios I have an installation that the Deploy Changes button is stuck and the discard wheel icon is hidden
13:33 angelberrios as per this commit in the code
13:33 angelberrios https://review.openstack.org/#/c/127436/1/nailgun/static/js/views/dialogs.jsx
13:34 angelberrios Originally was a 5.1 installation but I have upgraded to 5.1.1 (when the event happened) and yerterday because I found this, I did a V6 upgrade
13:34 angelberrios but still no discard changes button
13:35 angelberrios so this I gues my only option is to clear the cloumn at the DB level
13:35 angelberrios id | status      | name        | mode      | release_id | changes                                                                                                         | pending_release_id ---|-------------|-------------|-----------|------------|-----------------------------------------------------------------------------------------------------------------|------------------- 8  | operational | HLPRCLOUD01 | multinode | 2          | [{u'node_id': 1
13:35 angelberrios column changes content to blank
13:50 ppradhan joined #fuel
13:54 f1gjam BobBall :) glad it worked out
13:55 BobBall Ish.  I've still not got external access from the VMs but I think that's my problem now.
13:55 f1gjam oh
13:55 f1gjam external access from the VM's
13:55 f1gjam you setup yo network
13:55 f1gjam in horizon
13:55 f1gjam connect the router
13:55 f1gjam give the VM's a floating up
13:55 f1gjam IP
13:56 BobBall *nod* I know - but I meant via the healthcheck
13:56 BobBall which allocates floating IPs for you
13:56 BobBall Just a routing issue internally
13:56 f1gjam ah ok
13:56 BobBall I suspect my gateway for this subnet is broken
13:56 f1gjam most likely
14:15 admin0 joined #fuel
14:15 admin0 hi all .. how do I get to the screen to re-setup the fuel networking options
14:19 BobBall Wow - he really wanted to know the answer!
14:19 f1gjam you cannot re-setup once fuel is installed
14:19 f1gjam you need to re-install
14:19 f1gjam :D
14:19 BobBall He's already gone f1gjam
14:19 f1gjam LOL
14:19 BobBall within 2 minutes
14:19 f1gjam you know i should work for Mirantis UK :)
14:19 BobBall I imagine if you carry on answering questions they'll send you a t-shirt ;)
14:20 jskarpet joined #fuel
14:22 f1gjam LOL
14:22 f1gjam well they should be answering my questions :) im looking to get this setup for my company
14:33 Miouge_ joined #fuel
14:37 blahRus joined #fuel
14:43 coincoin joined #fuel
14:45 coincoin hi, I have deployed juno 6.0 with fuel on ubuntu 12.04, in HA
14:45 admin0 joined #fuel
14:45 admin0 my VM got rebooted after the install was finished, so now on reboot, fuel comes up with a differnet ip address and not the one that needs to be there
14:46 coincoin it works fine, but I was wondering about the ha option when creating a router : now that l3 -agent HA is handled by pacemaker, do I really need to create my router with the ha option
14:46 coincoin ?
14:47 claflico joined #fuel
14:50 subscope joined #fuel
14:50 admin0 joined #fuel
14:51 admin0 hello
14:55 bazilza joined #fuel
14:56 coincoin hello hello hello (annoy-o-tron mode on)
14:58 bazilza openstack is annoying enough
15:00 coincoin it may be fun, but it can also be annoying
15:00 coincoin Can someone help me with my ha option on router please ?
15:01 bazilza sadly, this week it is mostly annoying
15:01 coincoin :(
15:03 bazilza i need to go to production but oslo.messaging issues making it impossible. nova-compute do not reconnect back .. etc ... i'm tired :/
15:03 bazilza coincoin, you should ask. may someone will help
15:05 andriikolesnikov joined #fuel
15:05 coincoin I already ask my question, just before you come in the room
15:05 coincoin but I can reiterate for you ;)
15:05 coincoin t works fine, but I was wondering about the ha option when creating a router : now that l3 -agent HA is handled by pacemaker, do I really need to create my router with the ha option  ?
15:06 MiroslavAnashkin coincoin: Is it about routers inside OpenStack or about hardware router?
15:07 coincoin openstack router
15:07 coincoin but thinking to my question, I think I should use the ha option on router
15:09 bazilza not sure about the neutron --ha option. but pace l3-agent will create qrouter on another controller node in case of failure
15:11 bazilza also i think it's on by default
15:11 bazilza check l3_ha in neutron.conf
15:14 bazilza i have ha false and my qrouter namespaces recreated just fine, by neutron
15:14 bazilza pacemaker doesn't do it. it just makes sure l3-agents are cloned
15:23 Longgeek joined #fuel
15:31 adanin joined #fuel
15:34 mattgriffin joined #fuel
15:35 f1gjam bazilza / kaliya  - here is a diagram of my network setup - I have also sent this to support - it seems everyone is confused on how i should be able to access the admin/pxe network from external subnet. image is uploaded here: https://dl.dropboxusercontent.com/u/833212/Mirantis_Network.jpg
15:38 bazilza f1gjam, i might've misled you then. i meant you can assign IPs from public network to all the nodes. not the PXE network
15:38 coincoin @bazilza : thx for your advice, i'm thinking the same
15:39 kaliya sorry guys I've been a bit busy before
15:39 bazilza coincoin, fuel setup don't use l3_ha. i think. it uses distributed routing
15:40 f1gjam bazilza your saying i can assign public addresses to all nodes (physical) and use these addresses to access the nodes, but the Admin network is supposed to be not routable?
15:41 bazilza f1gjam, yes
15:42 f1gjam ok - i am going ot test that out now - rebuilding entire env
15:42 f1gjam if this is  the case, i will let support know and hope someone updates the docs to clarify :0
15:42 f1gjam :)
15:42 bazilza f1gjam, nothing stops you to make it routable though. you'll need to add your own routing though on the nodes or use a gateway machine in that network with NAT or somthing
15:44 f1gjam its setup like that
15:44 f1gjam but the installer configures them use the fuel master as GW
15:44 f1gjam and the controller loses it gateway completely
15:47 BobBall What do I need to do to add a custom DHCP interface to compute nodes?  Seems the firewall is blocking at least, but I don't understand why since the input chain includes DHCP acceptance and the OUTPUT chain just accepts everything.  Do I need to hack it in at the low-level or is there a higher level construct I can add to create a custom network for compute nodes?
15:49 bazilza f1gjam, controller must have ip in public network in order to route the traffic in/out and hosting floating ip rules. its default gateway should be public network usual gateway
15:49 f1gjam ok
15:51 e0ne joined #fuel
15:52 f1gjam support are now saying the only way to access the nodes
15:52 f1gjam is via the fuel master
15:52 f1gjam you cannot jump directly to the nodes
15:52 bazilza f1gjam, if you enable public IPs all nodes then default gateway will be the same as on controller
15:52 f1gjam this is how its decide
15:52 f1gjam sure
15:53 emagana joined #fuel
15:53 f1gjam bazilza  i think you are right, the documentation should clarify that
15:53 f1gjam and also state the role of fuel master has jump host
15:53 bazilza i have a setup with public ips on all nodes and it works just fine
15:54 f1gjam yes that would work but its not explained
15:54 f1gjam thats my point
15:54 f1gjam so ive spent like 2 days trying to figure out this routing issue
15:54 f1gjam and its not an issue its a design
15:54 bazilza f1gjam, may be. i don't read docs. i like pain :)
15:54 f1gjam thats why i thought :)
15:55 f1gjam *what
15:55 f1gjam but i changed two routers
15:55 f1gjam thinking they were messed up
15:55 f1gjam :S
15:57 bazilza f1gjam, i've spent last to days debugging neutron rabbitmq disconnects, and i still don't have a solution.
15:58 bazilza fuel guys saying they are working on porting oslo.messaging 1.8.0 to 6.0 ... hope it will happen soon
15:58 aarefiev joined #fuel
15:59 f1gjam bazilza thats a real issue my one isnt :)
15:59 bazilza until then i'm just retrying to create servers until meta-data service answers.. :/
16:03 bazilza f1gjam, well. you've got your answer now :) i hope to accumulate enough karma so someone else will help me ... :)
16:03 f1gjam LOL
16:03 f1gjam good luck
16:03 f1gjam you are a channel admin - i thought you worked for mirantis
16:04 bazilza i don't think i am
16:05 f1gjam woops not you kaliya
16:05 f1gjam :)
16:06 neophy joined #fuel
16:11 Salec joined #fuel
16:11 rmoe joined #fuel
16:17 f1gjam Bazila I can't select the option to assign the nodes public IP addresses
16:19 Salec left #fuel
16:33 f1gjam i found it
16:35 coincoin I think you can only tick the option if nothing is deployed in your envirionment
16:38 neophy Hi all, # cat /etc/resolv.conf
16:38 neophy nameserver
16:38 neophy , dns entry is missing on fuel master node. I am getting this http://paste.openstack.org/show/196668/ trigger from zabbix. I am not sure this issue in my environment or not. any help?
16:39 emagana joined #fuel
16:40 emagana joined #fuel
17:02 bazilza joined #fuel
17:08 xarses joined #fuel
17:13 admin0 joined #fuel
17:13 Longgeek joined #fuel
17:19 andriikolesnikov joined #fuel
17:35 emagana joined #fuel
17:47 martineg_ ppradhan: did you get an answer to your question earlier about slowdowns with keystone/dashboard after shutdown of one (of three) controller in HA mode?
17:48 martineg_ we're facing the same issue, I'm looking at https://bugs.launchpad.net/fuel/6.0.x/+bug/1405549 - looks more or less still unresolved?
18:06 MiroslavAnashkin martineg_: There is solution at the end of bug thread. https://bugs.launchpad.net/fuel/+bug/1405549/comments/22
18:07 okosse joined #fuel
18:16 jaypipes joined #fuel
18:17 xarses joined #fuel
18:26 mattgrif_ joined #fuel
18:27 martineg_ MiroslavAnashkin: ok thanks, was looking at this now .. will try this and re-run tests with shutdown of one controller.
18:32 jobewan joined #fuel
18:33 e0ne joined #fuel
18:55 muadgib joined #fuel
18:57 muadgib joined #fuel
19:24 Longgeek joined #fuel
19:39 emagana joined #fuel
19:55 mattgriffin joined #fuel
20:02 tzn joined #fuel
20:20 MRE joined #fuel
20:20 MRE How does Fuel deploy Ceph Metadata Servers? Ceph-MDS?
20:21 MRE does it put them on the monitor nodes, osd nodes. or is there a ceph-mds Role?
20:35 MRE Also, can Fuel deploy Nova pre-configured to run Docker? i"ll bet someone is working on that somewhere.
20:46 andriikolesnikov joined #fuel
20:54 aleksandr_null joined #fuel
21:03 tzn joined #fuel
21:13 Longgeek joined #fuel
21:27 mattgriffin joined #fuel
21:31 mattgrif_ joined #fuel
21:32 adanin joined #fuel
21:38 xarses joined #fuel
21:51 ppradhan left #fuel
22:08 admin0 joined #fuel
22:13 HeOS joined #fuel
22:19 jfluhmann joined #fuel
22:27 MRE ceph-mds is only for cephFS which we dont need for openstack. okay.
22:39 MRE unless docker under nova cant use RADOS, then cephFS would come in handy
22:43 e0ne joined #fuel
22:46 HeOS joined #fuel
22:54 MRE can nova - docker use ceph block device or does it need a file system?
23:08 HeOS joined #fuel
23:09 e0ne joined #fuel
23:10 f1gjam joined #fuel
23:10 Longgeek joined #fuel
23:39 xarses joined #fuel
23:41 HeOS joined #fuel
23:41 mescanef joined #fuel
23:59 tzn joined #fuel

| Channels | #fuel index | Today | | Search | Google Search | Plain-Text | summary