Earlier  
Posted Nick Remark
#openstack-nova - 2018-10-05
16:03:12 PapaOurs spatel: there was an excellent talk about AZs in a previous Summit
16:03:26 PapaOurs spatel: lemme help you by giving you the session link
16:04:00 spatel PapaOurs: great they must have video in youtube then
16:04:10 PapaOurs spatel: https://www.openstack.org/videos/vancouver-2018/curse-your-bones-availability-zones-1
16:04:25 spatel sweet!!
16:04:36 leakypipes fried_rice: :)
16:05:29 spatel trying to understand mechanic behind AZ and scheduler, how drive logic to spread machine out between two zone?
16:05:38 spatel is that scheduler job ?
16:06:24 PapaOurs spatel: that's a filter
16:06:35 PapaOurs but see the video, it explains it :)
16:06:43 spatel i am watching it..
16:06:50 spatel PapaOurs: thanks
16:08:15 PapaOurs 6pm here, time for opening a beer
16:08:21 PapaOurs have a nice week-end folks
16:08:23 PapaOurs \o
16:09:18 sean-k-mooney spatel: sorry just looking at teh diagram now
16:10:12 spatel no worry! take your time
16:10:38 spatel SR-IOV doesn't support bonding so i can't create bond between two TOR switches
16:11:05 spatel SR-IOV does support bonding but inside VM (i think i am not ready for that)
16:11:35 sean-k-mooney so availablity zones are a user facing grouping mechaniums. from the schduler point of view they are just a metadata tag on a host aggaragte
16:12:18 sean-k-mooney so if you requst a vm in zone A the scudler will only look at host that are in a host aggragte with avaiablity_zone=a set
16:13:05 sean-k-mooney it then allpies the rest of the filters to reduce the set of host. e.g. ram, cpus, disk or in your case pci deivces for sriov
16:13:10 spatel but that would be hardwire right?
16:13:40 sean-k-mooney that is a bit of a simplicfation as there is another service called placement that is also invoved.
16:13:46 sean-k-mooney so yes and no
16:14:03 spatel currently we are using AWS with terraform and we tell tf to create instance between A and B zone and it does it job
16:14:10 sean-k-mooney the operator e.g. you in this case has to decalre what hosts are in the A avaliablity zone
16:14:32 sean-k-mooney spatel: AWS availablity zones ar enot the same as openstacks
16:15:03 spatel understand but is there way i can get functionality close to AWS
16:15:08 sean-k-mooney in AWS an availablity zone is a seperate falut domain so if you have an ouate in zone A zone by is still fine
16:16:25 spatel Let's in openstack i create two AZ (a, b) and i want to tell openstack go build 10 vm then does openstack spread machine out between two AZ ? or this is not possible?
16:16:44 sean-k-mooney spatel: yes unfortunetely we leave that as an exerciest to the operator to configure cooreltys. 1.) you can actully map your AZ to seprate falut domains 2.) you can use cells + availahity zone or 3.) you can use regions
16:16:57 spatel if not possible then i have to specify AZ by hand
16:17:51 sean-k-mooney spatel: if they are different requires e.g. 10 vm requires they can be spread. if you use the multi create api im not certine if it will spread or scdule them all to the same AZ
16:18:08 spatel currently i have TOR level AZ (i am not much worried about power failure)
16:18:54 spatel just want to make sure if i loose TOR it won't knock out all my customer server
16:19:15 spatel so trying to spread out VM between two tor
16:19:18 sean-k-mooney ok the other thing that you are sharing with openstack AZ are your controler service. e.g. your db and api seriveces
16:19:51 spatel hmm??
16:20:47 sean-k-mooney so in a standard openstack deployment you have contoler nodes that run the api services, schduler and databases
16:21:03 spatel Yes i have 3 controller node
16:21:15 sean-k-mooney thoes contoll plane nodes manage the compute nodes that reside in AZs
16:22:11 spatel yup
16:22:29 sean-k-mooney yes so in your 2 AZ config as long as the compute node can sitll acess one of the 3 controllers and you can also acess the same controler then you will be able to manage that host
16:23:12 sean-k-mooney if you have a network partition that breaks that link then the guests workload running on the compute nodes may still work fine but you will lose the ablity to manage them
16:23:57 spatel i am totally fine if i loose management access!! as far as application serving my customer
16:24:17 spatel replacing TOR switch will take may be 1 hour or so!!
16:24:17 sean-k-mooney in aws the api/dataables/storage and networking uplinks are deployed per AZ so the failrue more for openstack is slightly different as the contolplane is not per aZ
16:24:37 spatel as far as i meet my customer SLA we are fine with that
16:25:45 sean-k-mooney ok :) many people coming form AWS expect openstacks availablity zone to map exactly to the same concept as AWS. you seam to have done more reasearch or given it more taught then most
16:25:46 spatel In my design if you see i break total server between two TOR switch and if either one down i will loose 50% machine
16:26:14 spatel I totally get that point AWS != openstack in AZ context :)
16:26:38 sean-k-mooney spatel: so looking at https://ibb.co/j9kgUK can you discibe your intient
16:26:53 sean-k-mooney you have 2 tors
16:26:58 spatel yes
16:27:06 sean-k-mooney and 2 AZs one per tor
16:27:10 spatel yes
16:27:38 sean-k-mooney half of your servers use TOR-1 for there SRIOV dataplane and TOR-2 for managmente
16:27:52 spatel yup! that is correct
16:28:08 sean-k-mooney the other half swap which TOR is used form management and data
16:28:22 spatel yup!
16:29:07 sean-k-mooney so if TOR-1's PSU explodes half the servers will loose data conenctivity and the other half will loose manament capablity
16:29:21 spatel yes! that is right
16:29:37 sean-k-mooney am is that what you want?
16:30:19 spatel yes in PSU failure still my 50% instance will be live and providing service but openstack will loose management function
16:30:20 sean-k-mooney i assume you do not have the option to seperate management and data traffic
16:31:36 sean-k-mooney spatel: do you intend to pass the PF or VF to the vms running your voip applications?
16:31:36 spatel I don't have option because i have to purchase NIC to install on compute node also SR-IOV doesn't do bonding so i am going to loose some customer if TOR switch blowup
16:32:10 spatel yes my VF is attach to instance
16:32:31 spatel i don't have any kind of bonding anywhere at present because of limitation of SR-IOV
16:33:02 spatel http://paste.openstack.org/show/731595/
16:33:24 spatel This is what my compute look, i am running 2 instance with dual NIC on two different VLAN
16:34:31 sean-k-mooney right so what i was goint to suggest is that you could either reserve a VF or use the pf of your sriov nic as a bond pair for your manament network.
16:35:34 spatel oh!! wait is that possible?
16:35:52 sean-k-mooney so in your case you can create a bound between eno1 and eno2 for managment traffic and then openstack can use just the vfs
16:36:00 sean-k-mooney spatel: yes
16:36:01 spatel i thought if you configure NIC for sr-iov then you can't bond them
16:36:13 sean-k-mooney it depends on the nic
16:36:36 spatel I have tried that and found my loop in my switch
16:36:44 spatel may be it was config issue..
16:37:18 spatel i have all Cisco nexus switch running vPC (active+active bonding mode=4)
16:37:33 sean-k-mooney well it defintely depens on how you create the bond (e.g. just mac loadbalcing with out swich supprot or lacp with it)
16:37:33 spatel i have Qlogic NIC
16:38:41 spatel did you every run that kind of test?
16:38:59 spatel i would like to know if you did and what config you configure
16:39:13 spatel I will run same experiment on my LAB and try come combination
16:39:33 spatel that was great if it work
16:39:42 sean-k-mooney i have never used Qlogic NIC but i have created bonds with intel nics in active failover mode
16:39:47 spatel s/was/would be/
16:40:06 spatel did you configure with SR-IOV ?
16:40:24 spatel i meant SR-IOV + Mgmt (in bonding)
16:40:47 spatel I think i understand what you trying to explain.. i will give it a try..
16:41:13 spatel i wish if you can create bonding between two VF :)
16:41:17 sean-k-mooney that is a good questiong the host had both sriov and a manament bound but i cant remeber if the manament bound was using the pf that had the sriov vf allocated from it
16:42:01 spatel i remember last time when i try to did that i got interface error (BPDU guard disable my nic)
16:42:10 sean-k-mooney i think i had 10G nic for tenat data/vfs 1 1G nic and a second 10G nic bonded for manament and ceph traffic
16:42:33 spatel oh!! you have PF level isolation
16:42:57 spatel I have plan in future to install 1G nic for Mgmt traffic but not this year
16:43:10 sean-k-mooney spatel: yes i did in that case but im tring to remeber if i tried to create a bound with the pf used for the VFs
16:44:05 spatel hmm! worth running some test :)

Earlier   Later