----- Original Message ----- From: "Michal Fojtik" <[email protected]> To: [email protected] Sent: Wednesday, December 15, 2010 6:35:13 PM GMT +10:00 Brisbane Subject: Re: Initial import of new RHEV-M driver
On 14/12/10 07:44 +1000, Stephen Gordon wrote: >On 12/14/2010 03:28 AM, [email protected] wrote: >>Hi guys, >> >>I spend most of this day and yesterday to get new RHEV-M API running. >>Actually I'm able to run Mock API, which is not perfect but for now >>enough to start working on this driver. >> >>So now what is working and what is not: >> >>+ Listing of realms (using DataCenter) (/api/realms) >> I'm using Datacenters instead of StorageDomains, because it has >> more sense to me, let me know what do you think. >> >>+ Listing of templates (/api/images) >> Basically there are all needed informations for this collection >> already in place, except 'owner_id', so using 'self' here for now >> should fix it. >> >>+ Hardware profiles >> OK, so nwo I'm not that sure ;-) I found out that there are basically >> two types of templates/VMs. One is DESKTOP and second SERVER. >> So I used these as a names for hardware_profiles. >> For memory I used a range from 1024..32*1024. I need to know what is >> the minimum amount of RAM for new VM under RHEV. Also reporting >> how much memory hypervisor provide would be appreciated here. >> As an architecture I used 'x86_64', storage is opaque. > >I believe in the current edition (RHEV 2.2) the minimum amount of RAM >that can be assigned to a guest is 512mb. The maximum that can be >assigned to a 64 bit guest appears to be listed as 265 GB. See: [SG] That should of course have been 256 GB. :) >http://docs.redhat.com/docs/en-US/Red_Hat_Enterprise_Virtualization_for_Servers/2.2/html/5.5-2.2_Hypervisor_Deployment_Guide/chap-Deployment_Guide-Requirements_and_limitations_of_Red_Hat_Enterprise_Virtualization_Hypervisors.html Thanks for this information, it's extra usefull for me! -- Michal > >>+ Instance states >> I used EC2 state-machine, which is very similar to RHEV-M one. >> In RHEV there is a lot of additional states, which could be >> 'simplified' to our case of PENDING, STOPPED, RUNNING >> >>- Instances >> Almost every properties are mapped here, like name, state, id... >> There I ran into problems with mock driver, because in mock you don't >> have 'cluster' element, which refers to cluster which refers to datacenter. >> So mapping hardware profile to instance is in this case not possible. >> Anyway, I tryied to follow documentation here so it should work, but I >> can't >> test it live.... >> >>Also few notes of rhev_client, which is a small library/class for connecting >>to RHEV-M API and calling operations here. >>I used 'method_missing' to map all resources, otherwise it will require a lot >>of repeated code in this client which will not look very nice. >> >>Let me know what do you think, all suggestions, questions are more than >>welcomed. >> >>PS: Looking for some real testing RHEV-M instance which I can use for real >>testing. >> I offering a good Czech beer for that ;-) >> >> -- Michal >> >> > -- -------------------------------------------------------- Michal Fojtik, [email protected] Deltacloud API: http://deltacloud.org --------------------------------------------------------
