Perl 6 - the future is here, just unevenly distributed

IRC log for #fuel, 2015-02-02

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

All times shown according to UTC.

Time Nick Message
01:07 LiJiansheng joined #fuel
02:58 dhblaz joined #fuel
03:04 Longgeek joined #fuel
03:04 Longgeek joined #fuel
04:31 dhblaz joined #fuel
04:34 Longgeek joined #fuel
05:32 Longgeek joined #fuel
05:32 jobewan joined #fuel
06:14 ricolin joined #fuel
06:34 teran joined #fuel
06:48 sambork joined #fuel
06:49 monester_laptop joined #fuel
07:05 artem_panchenko left #fuel
07:05 artem_panchenko joined #fuel
07:19 corepb joined #fuel
07:22 dklepikov joined #fuel
07:30 Longgeek joined #fuel
07:35 kozhukalov joined #fuel
07:40 sambork joined #fuel
07:47 sovsianikov_ left #fuel
07:47 Miouge joined #fuel
07:49 Longgeek joined #fuel
08:08 sovsianikov_ joined #fuel
08:09 sovsianikov_ left #fuel
08:09 sovsianikov_ joined #fuel
08:09 Miouge_ joined #fuel
08:09 sovsianikov_ left #fuel
08:11 sovsianikov_ joined #fuel
08:30 hyperbaba joined #fuel
08:30 okosse joined #fuel
08:31 sovsianikov_ left #fuel
08:32 sovsianikov_ joined #fuel
08:34 sovsianikov_ left #fuel
08:34 sovsianikov_ joined #fuel
08:35 sovsianikov_ left #fuel
08:36 ricolin joined #fuel
08:39 sovsianikov_ joined #fuel
08:39 rtau joined #fuel
08:39 ebogdanov joined #fuel
08:44 ebogdano_ joined #fuel
08:51 stamak joined #fuel
09:01 mkwiek joined #fuel
09:01 vks joined #fuel
09:03 azemlyanov joined #fuel
09:06 vks i have installed a multi node HA setup using Mirantis fuel. I see haproxy namespace on all nodes. In that i see "hapr-ns" interface on all nodes. What is the role of that interface?
09:18 dklepikov vks: Hello. What fuel version do you have?
09:21 kivio joined #fuel
09:21 vks dklepikov, 5.1.1
09:21 vks dklepikov, hi
09:24 mescanef joined #fuel
09:29 vks dklepikov, is there any doc or such which explains the working of Haproxy in openstack HA?
09:29 subscope joined #fuel
09:30 alecv joined #fuel
09:35 devstok joined #fuel
09:40 devstok HI guys
09:40 dklepikov vks: http://docs.mirantis.com/openstack/fuel/fuel-6.0/reference-architecture.html?highlight=haproxy#openstack-environment-architecture
09:40 devstok I'm always stopped here
09:40 devstok (Puppet::Type::Neutron_floatingip_pool::ProviderNeutron) Neutron API not avalaible. Wait up to 42 sec.
09:40 devstok when this ends... gives a timeout error of neutron API
09:41 devstok first controller installed but the second and third go in timeout
09:41 dklepikov vks: http://docs.mirantis.com/openstack/fuel/fuel-5.1/reference-architecture.html#multi-node-with-ha-deployment
09:41 avgoor joined #fuel
09:44 vks dklepikov, i see that
09:47 vks dklepikov,  i have 3 interfaces in haproxy namespace, "hapr-p", "hapr-ns", "hapr-m". I do not understand why "hapr-ns"?
09:47 hyperbaba Hi, i've updated fuel to 5.1.1 from 5.1. Now dns resolving on 10.20.0.2 interface is not working rendering my controller nodes inoperable. How can i fix this?
09:47 dklepikov vks: here is a script https://github.com/stackforge/fuel-library/blob/master/deployment/puppet/cluster/files/ocf/ns_haproxy
09:48 dklepikov vks: Description:  Manages haproxy daemon as an OCF resource in an High Availability setup inside a namespace
09:55 ddmitriev joined #fuel
09:58 kaliya joined #fuel
10:09 teran joined #fuel
10:28 vks dklepikov, http://pastebin.com/6MnGxfrt
10:29 devstok why my deployment goes in timeout
10:32 t_dmitry joined #fuel
10:35 e0ne joined #fuel
10:35 dklepikov vks: Every OpenStack controller runs HAProxy, which manages a single External Virtual IP (VIP) for all controller nodes and provides HTTP and TCP load balancing of requests going to OpenStack API services, RabbitMQ, and MySQL.
10:35 dklepikov When an end user accesses the OpenStack cloud using Horizon or makes a request to the REST API for services such as nova-api, glance-api, keystone-api, neutron-api, nova-scheduler, MySQL or RabbitMQ, the request goes to the live controller node currently holding the External VIP, and the connection gets terminated by HAProxy. When the next request comes in, HAProxy handles it, and may send it to the original controller or another in the environment, de
10:35 dklepikov pending on load conditions.
10:35 vks dklepikov, correct
10:36 geekinutah joined #fuel
10:39 teran joined #fuel
10:41 teran_ joined #fuel
11:11 vks dklepikov, the request always first lands on HAProxy external IP and then it get forwarded to any controller node. What does that mean: "When the next request comes in, HAProxy handles it"?
11:18 dklepikov vks : HAProxy decides on what controller send request, depending on load conditions.
11:20 vks dklepikov, thats true, but didn't understand "the next request HAProxy handles ", isn't it that first request also go to haproxy only
11:22 dklepikov vks : all requests going via HAProxy
11:23 dklepikov vks : http://docs.mirantis.com/openstack/fuel/fuel-5.1/reference-architecture.html#ha-logical-setup
11:23 dklepikov vks : http://docs.mirantis.com/openstack/fuel/fuel-5.1/reference-architecture.html#details-of-multi-node-with-ha-deployment
11:59 Miouge joined #fuel
12:10 Miouge joined #fuel
12:31 subscope joined #fuel
13:23 mkwiek joined #fuel
13:25 corepb joined #fuel
13:32 sambork joined #fuel
13:39 adanin joined #fuel
13:50 Miouge After a cluster install (Ubuntu install went through) I server is stuck at PXE boot (see http://i.imgur.com/odpY6k6.png )
13:54 Miouge Mmm it is booting after removing all drivers and replugging them, probably something not attached correctly
13:58 Miouge *all drives
14:14 mescanef joined #fuel
14:30 jaypipes joined #fuel
14:49 aarzhanov joined #fuel
14:50 aarzhanov left #fuel
14:51 aarzhanov joined #fuel
14:51 aarzhanov left #fuel
14:55 aarzhanov joined #fuel
14:56 aarzhanov left #fuel
14:57 corepb joined #fuel
14:58 aarzhanov joined #fuel
15:05 miroslav_ joined #fuel
15:06 teran joined #fuel
15:09 mpetason joined #fuel
15:13 blahRus joined #fuel
15:15 jobewan joined #fuel
15:19 aarzhanov joined #fuel
16:10 aarzhanov joined #fuel
16:10 aarzhanov left #fuel
16:20 e0ne_ joined #fuel
16:38 monester_laptop joined #fuel
16:49 teran_ joined #fuel
17:18 mattgriffin joined #fuel
17:34 vks joined #fuel
17:40 angdraug joined #fuel
17:47 rmoe joined #fuel
17:47 xarses joined #fuel
17:51 vks I have installed openstack multinode HA with help of fuel.  can somebody explain me the working of two VIPs in haproxy namespace?
17:54 moizarif joined #fuel
17:55 moizarif hi all,my node-id on fuel master has gone upto 90. How can i reset this count back to 1 ?
17:57 MiroslavAnashkin vks: One VIP is for internal (management) network, one for external (public)
17:57 MiroslavAnashkin These VIPs are the single entry points for HA services
17:58 MiroslavAnashkin moizarif: Do not do it. It is not easy and possible only if you have no one node in Fuel database.
17:59 vks MiroslavAnashkin, thats true, all the openstack services gets binded on external vip , but see the request goes via internal network. why so?
17:59 MiroslavAnashkin Tenant networks need to reach out Neutron - which is scattered across all controllers
18:01 MiroslavAnashkin Internal VIP is to connect one of Neutron instance and neutron agents on the compute nodes
18:02 vks i see it for all the services nova, neutron , glance etc.
18:04 MiroslavAnashkin Yes, Neutron virtual networks exist inside the internal network and all need some way to communicate with single Neutron service instance of several available.
18:05 e0ne_ joined #fuel
18:06 moizarif MiroslavAnashkin: i have deleted all my nodes so have anything in my fuel database at the moment. How can i reset the count now?
18:07 moizarif nothing in my fuel database**
18:15 xarses moizarif: you want to reset the nodes couter?
18:16 xarses the easiest way is to reload the db if you already destroyed all of the clusters and nodes
18:18 moizarif @xarses: yes i want to reset the nodes counter
18:20 teran joined #fuel
18:20 vks joined #fuel
18:23 xarses if you want to reset the database (again this will destroy everything, clusters and all) then you can follow https://gist.github.com/xarses/10523d86f59788d11e75
18:23 vks left #fuel
18:23 xarses if you want to reset just the counter, then you need to go into the postgres db and reset it
18:23 vks joined #fuel
18:24 vks MiroslavAnashkin, ??
18:26 xarses vks, it's around dinner time for MiroslavAnashkin so he probably stepped away
18:26 vks ok np
18:26 vks xarses, can u answer? correct me if i am wrong. all the request come as destination of External VIP. Haproxy change the dst to one of the servers configured on internal vip
18:27 vks why external interface is configured as veth pair?
18:28 xarses haproxy runs in a network namespace for both external (public) and management vips
18:28 vks correct
18:28 xarses haproxy in the namespace will redirect the request onto the management network for the controllers
18:28 vks correct it changes the dst ip??
18:29 xarses it probably creates a new packet, but effectively it changes the dst ip
18:29 vks i see external (public) and management as veth pair
18:29 vks for mgmt its to send the pkt out
18:29 vks but why for external
18:30 vks ?
18:32 vks also there are multiple neutron servers and neutron agents on single node, how it works?
18:32 xarses I'd have to double check, but i think the veth pairs on the external are required for the traffic to properly traverse the namespace
18:33 xarses can you send a paste of that you want clarified so i can get back to you when I have a controller up
18:34 xarses vks: for neutron server, its stateless so it's the same as any other web app, you put any of them behind a lb and it scales well. Messages are placed on amqp bus and handled by other components
18:35 vks yeah but how does mutiple server gets binded on single port?
18:37 vks xarses, http://pastebin.com/u6jRfqXw
18:38 omolchanov joined #fuel
18:48 kupo24z joined #fuel
18:52 adanin joined #fuel
18:53 xarses joined #fuel
18:55 angdraug joined #fuel
19:05 vks joined #fuel
19:10 xarses for neutron-agent there is usually only one on each node (fuel does this) but there are ways in openstack that this is allowed with some configuration
19:11 xarses vks: ^
19:16 thumpba joined #fuel
19:25 teran joined #fuel
19:49 teran joined #fuel
19:52 ddmitriev joined #fuel
20:29 MiroslavAnashkin BTW, we run HAProxy inside the namespace as well. So, we need a way to connect HAProxy with VIPs. Veth work better than patch. And it saves HAProxy from going down if VIP goes down.
20:53 angdraug joined #fuel
21:01 dhblaz joined #fuel
21:02 dhblaz I have a fuel 4.0 Centos install that I need to upgrade glibc for protection against GHOST.  Can someone recommend where I get glibc rpms for that?
21:26 Miouge joined #fuel
21:48 rubick_ joined #fuel
22:44 e0ne joined #fuel
22:49 angdraug dhblaz: same glibc that we've published for 6.0 should work the same with 4.0
22:49 angdraug it's the same centos version after all
22:50 dhblaz Is there a support article about this upgrade?  Or where is the rpm published?
22:50 angdraug dhblaz: https://software.mirantis.com/mirantis-openstack-technical-bulletins/
22:50 dhblaz Thank you
22:52 angdraug you're welcome:)
23:51 SergK joined #fuel
23:51 mattgriffin joined #fuel

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