@olivier here is the opposite side: an example of the ovirt->libvirt communication channel

@olivier YUP! and the http server is plain httprequest stuff from CPython, http framework around or stuff like uwsgi!

@olivier YUP! it seems I was wrong. Red Hat has evetually rewritten it! In Python!

@olivier read here.

it was 2008 citrix acquired xen. red hat reacted buying the only other solution on the market. kumranet were strong in pushing their own orchestration solution.

part of it (the core: KVM) was open. tools givin added value (orchestrator) were closed and billed. don't know why but kumranet written everything in java. Maybe faster to develop then c++. less crosscompile (think mainframes). maybe node or python where simply too young or they where not happy with python speed - node was young for sure.

Red hat gots the entire blob. they have reworked it not rewritten.

@jaredbusch said in What is KVM Best Management Tools in 2017?:

@matteo-nunziati said in What is KVM Best Management Tools in 2017?:

https://access.redhat.com/documentation/en-US/Red_Hat_Enterprise_Virtualization/3.6/pdf/REST_API_Guide/Red_Hat_Enterprise_Virtualization-3.6-REST_API_Guide-en-US.pdf

still I do not find the opened source of RHEV

Who said that RH had to open source everything?

Not they have to. Simply they have worked a lot to open source their virtualization infrastructure. Ovirt was born by the rewriting of the original orchestration tool written by the KVM company.

@matteo-nunziati said in [What is KVM Best Management Tools in 2017?]> maybe this ovirt page refers to the same but you just need the api. if ovist is the source then you had to deal with the http wrapper around VDSM on top of libvirt on top of qemu/kvm.

How else would you access the API without http? Something has to listen.

http is not the issue here. VDSM is. very complex beast. written in java.

kimchi has always been a single server solution. In the beginning it was ubrellaed (<-?WTF?) by ovirt as web gui for their nodes. eventually it has been replaced by cockpit.

https://access.redhat.com/documentation/en-US/Red_Hat_Enterprise_Virtualization/3.6/pdf/REST_API_Guide/Red_Hat_Enterprise_Virtualization-3.6-REST_API_Guide-en-US.pdf

still I do not find the opened source of RHEV

maybe this ovirt page refers to the same but you just need the api. if ovist is the source then you had to deal with the http wrapper around VDSM on top of libvirt on top of qemu/kvm.

@storageninja said in Veeam Backup and Replication for ESXi:

@tim_g said in Veeam Backup and Replication for ESXi:

I don't hate Veeam. I just hate how they support some of their products.

They will support it, you are just not willing to pay....

I think what @Tim_G says is that he has the perception that Veeam releases the linux agent for free but then doesn't care about maintaining it compatible or giving proper support in the community . He probably refers to the difference found in support forums between a free commercial product and a proper opensource product with its own community. Two very different beasts IMHO.

I've just tried their agent for windows once and never looked in it soo much, therefore can't say.

The only think I know is that breaking a linux machine is quite hard to me and I tend to use stuff like rsnapshot for my spare time stuff... and a clonezilla just a beginning of a new deployment (just in case). All my SMB workloads are VM running agentless backups with Altaro at HV level. And my linux VM are all deployed with Ansible now so I really don't mind so much about more complex backup solutions for Linux.

I mostly mind for windows... 'coz it really scarries me :-s !

@dashrender noob question: which is the equivalent of cat /etc/version on win powershell?!

Never got the cord. Always found news to be manipulated. Stop watching at them. Found cultural stuff fading away from tv. Basically Ive found myself switching tv on only for searching for good movies. when I was ready, not when they were. Put the tv in the trashcan, started pay netflix.

@travisdh1 said in What was your first Linux/Unix distro?:

How did I miss this thread?

IRIX - SGI Equipment, 1996

1x Origin 2000 Server
4x 64-bit CPU,
1GB RAM,
8x300GB 15k SCSI HDD

Should I be concerned that I still remember the server specs?

4x Octane Workstations

2x O2 Workstations

Finally, 4x Indigo2 Workstations

It was expensive, but man they made good looking kit!

My bet was on you and irix or similar

@jaredbusch said in Spec me a new desktop:

@scottalanmiller said in Spec me a new desktop:

@jaredbusch said in Spec me a new desktop:

With the Intel NUC does anyone know if Fedora works well there?

I would miss my 4TB 3.5" SATA drive, but I can buy a large 2.5" drive.

@scottalanmiller what about those HP boxes you linked. Any idea how well Fedora supports the hardware?

No idea, I've never tried on those. I've never used those at all, they just look really nice.

They do, that is why I am curious.

Until you try no one knows but as a general rule consider intel as highly compatibile briadcom as a mixed bag anything else as a no go...

@tim_g said in Microsoft Removes Storage Spaces Direct from Windows Server 2016:

You can still deploy systems with S2D by using RS1, and is still supported as in LTS.

holy shit, please can you enlight me on how MS numbers/support win versions?! I'm going mad. I've hyper-v server 2016 since march, always installed only security patches: which f***ing version should I expect?!

Also have some win server 2016 licenses... how to"stick" on RS1 install and what RS1 is?

X-(

BTW: S2D also included the LVM? or this is "just" storage space and is still available.
Hell, linux is way more straight...

first computer I used: Philips MSX

first pc I owned was a withe box w/ Pentium I MMX.
In the middle I played with Commodore Amiga 500

@dashrender slackware linux don't remember the version. First comment was:WTF. Second one was: cool you can recompile the os. Third was: hei I was used to have a modem...

ok,

I'm still confused about the terminology. Please, can you confirm and/or fix my interpretation?

1. critical: not a security fix, but something seriously broken is fixed here.
2. security: a security fix, something seriously broken is fixed here and a security hole is closed.
3. cumulative: just use this if you have missed a lot of stuff and you have a long queue updates otherwise. Download this bulk huge fatty thing and align with upstream.
4. all other types... really too many types here, do they really worth?

Now the second question.

Having an hyperv server 2016 should I need to install critical fixes? I mean: if I do not notice any misfunction should I patch? to me it is right to patch as you never know when a critical bug will hit your usage.

Also, as a general rule I patch manually and I always require recommended updates only. Then I choose what to install.

any opinion/correction/suggestion?!

thank you.

@scottalanmiller said in how to assign vCPU and memory on virtual machines on VMware vSphere?:

@travisdh1 said in how to assign vCPU and memory on virtual machines on VMware vSphere?:

@scottalanmiller said in how to assign vCPU and memory on virtual machines on VMware vSphere?:

@travisdh1 said in how to assign vCPU and memory on virtual machines on VMware vSphere?:

@ghani Assuming each host is running, that likely will not be an issue. If one host goes down, they will likely have huge contention issues. You'd have to expect that with a single 48 cpu cores assigned 125 virtual cores.

Without seeing any statistics on performance, I'd say they're ok currently. They could probably downsize the number of cpu cores assigned to many of the virtual machines if they do start running into cpu contention issues.

Even then for typical workloads would be just fine.

True, people (me included), do tend to assign way more vcpu than actually needed.

I think the answer to this question is: no, but I'll ask anyway. Is there a good standard with the over provisioning of cpu in virtual environments? (IE: 2.6:1 like we'd see in a server failure here.)

I have a feeling that even 4:1 isn't too bad in most cases.

we are currently at 8:1

@scottalanmiller about RDS. I'm still trying to understand the licence costs. you need RPC cals AND standard cals?! or is it just one? also how do you pay for apps, like office?! just enter the account per user and get your office 365 licence or is it way more complex/expensive?!

well. I've started considering VDI even for my current company: aroud 42 seats, expanding.

But then I've started condisering: why VDI rather than RDP services?! still don't know the answer, because I've not digged in enough, but, mind that you can also do RDP services before VDI.

And I'm talking windows, not linux.

@storageninja said in Xenserver and Storage:

@olivier said in Xenserver and Storage:

That's why I asked if you have better knowledge of community on this solution because I really don't. So if it's the case, that it's not stable (darn, it's here since a long time!), that's indeed not an option.

Issue with HA lizzard is that it doesn't have a stateful quorum system (just pinging a single IP address). You can split brain it.

Afaik latest ha-lizzard docs suggest to work in active-passive only... There must be a reason...