Re: [ovirt-users] fileserver as a guest oVirt
--_000_5F9E965F5A80BC468BE5F40576769F098C555F4Cexchange21_ Content-Type: text/plain; charset="utf-8" Content-Transfer-Encoding: base64 DQpEZW4gMTIganVsIDIwMTQgMjI6NDkgc2tyZXYgTmlrbGFzIEZvbmRiZXJnIDxuaWtsYXNAdmly ZW9uZS5jb20+Og0KPg0KPg0KPg0KPiBPbiAxMiBqdWwgMjAxNCwgYXQgMTY6NTcsICJLYXJsaSBT asO2YmVyZyIgPEthcmxpLlNqb2JlcmdAc2x1LnNlPiB3cm90ZToNCj4NCj4+DQo+PiBEZW4gMTIg anVsIDIwMTQgMTU6NDUgc2tyZXYgTmlrbGFzIEZvbmRiZXJnIDxuaWtsYXNAdmlyZW9uZS5jb20+ Og0KPj4gPg0KPj4gPiBIaSwNCj4+ID4NCj4+ID4gSeKAmW0gbmV3IHRvIG9WaXJ0IGJ1dCBJIG11 c3Qgc2F5IEkgYW0gaW1wcmVzc2VkIQ0KPj4gPiBJIGFtIHJ1bm5pbmcgaXQgb24gYSBIUCBETDM4 MCB3aXRoIGFuIGV4dGVybmFsIFNBUyBjaGFzc2kuDQo+PiA+IExpbnV4IGRpc3QgaXMgQ2VudG9z IDYuNSBhbmQgb1ZpcnQgaXMgMy40IHJ1bm5pbmcgYWxsLWluLW9uZSAoZm9yIG5vdyB1bnRpbCB3 ZSBuZWVkIHRvIGhhdmUgYSBzZWNvbmQgaG9zdCkuDQo+PiA+DQo+PiA+IE91ciBjb21wYW55ICh3 d3cudmlyZW9uZS5jb20pIGRlYWxzIHdpdGggc3lzdGVtIGFyY2hpdGVjdHVyZSBmb3IgbWFueSB0 ZWxjbyBhbmQgbWVkaWEgb3BlcmF0b3JzIGFuZCBpcyBub3cgc2V0dGluZyB1cCBhIHNtYWxsIG93 biBkYXRhY2VudGVyIGZvciBvdXIgaW50ZXJuYWwgdGVzdHMgYXMgd2VsbCBhcyBvdXIgSVQgaW5m cmFzdHJ1Y3R1cmUuDQo+PiA+IFdlIGFyZSBpbiB0aGUgcHJvY2VzcyBvZiBpbnN0YWxsaW5nIFpl bnR5YWwgZm9yIHRoZSBTTUIgcHVycG9zZXMgb24gYSBndWVzdCBhbmQgaXQgd291bGQgYmUgZ3Jl YXQgdG8gaGF2ZSB0aGF0IGd1ZXN0IGFsc28gc2VydmluZyBhIGZzIHBhdGggZGlyZWN0b3J5IHdp dGggTkZTICsgU01CICh3aGljaCBpcyBzZW1pIGNyaXBwbGVkIG9uIHRoZSBob3N0IGFmdGVyIG9W aXJ0IGluc3RhbGxhdGlvbiB3aXRoIHZlcnNpb24gMyBldC5jLikuDQo+PiA+DQo+PiA+IERvZXMg YW55b25lIGhhdmUgYW4gaWRlYSBvZiBob3cgSSBjYW4gdGhyb3VnaCBvVmlydCAoc2VlbiBzZXZl cmFsIHNvbHV0aW9ucyB1c2luZyB2aXJzaCBhbmQga3ZtKSBsZXR0aW5nIG15IFplbnR5YWwgVWJ1 bnR1IGd1ZXN0IGhhdmUgYWNjZXNzIHRvIGEgaG9zdCBtb3VudCBwb2ludCBvciBpZiBuZWNlc3Nh cnkgKHNlY29uZCBiZXN0KSBhIHNlcGVyYXRlIHBhcnRpdGlvbj8NCj4+ID4NCj4+ID4gQmVzdCBy ZWdhcmRzDQo+PiA+IE5pa2xhcw0KPj4gPg0KPj4NCj4+IFdoeSBub3QganVzdCBnaXZlIHRoZSBn dWVzdCBhIHRoaW4gcHJvdmlzaW9uIHZpcnR1YWwgaGFyZCBkcml2ZSBhbmQgZXhwYW5kIGl0IG9u IGEgZGVtYW5kIGJhc2lzPw0KPj4NCj4+IC9LDQo+DQo+IFRoYW5rcyBmb3IgdGhlIGFkdmlzZSBi dXQgdGhpcyB3b3VsZCBub3Qgc3VpdGUgdXMgSSdtIGFmcmFpZC4gSXQgd291bGQgYmUgZGlmZmlj dWx0IHdydCBpbmNyZW1lbnRhbCBiYWNrdXBzIGFzIHdlbGwgYXMgaG9zdCBtYWNoaW5lIGZpbGUt cm91dGluZXMuDQo+DQo+DQoNCldlbGwsIGdvaW5nIGJ5IE9jY2FtJ3MgcmFpem9yOyB0aGUgc2lt cGxlc3QgYW5zd2VyIGlzIHVzdWFsbHkgY29ycmVjdC4gQ2FuJ3QgcmVhbGx5IHRlbGwgd2hhdCB5 b3UgbWVhbiBieSBmaWxlLXJvdXRpbmVzIGJ1dCBiYWNrdXBzIHdvdWxkIGJlIHdlbGwgc2VydmVk IGJ5IHNuYXBzaG90cyAoY2FuJ3QgZ2V0IG1vcmUgaW5jcmVtZW50YWwgdGhhbiB0aGF0KSBhbmQg ZGlzYXN0ZXIgcmVjb3ZlcnkgY291bGQgYmUgYXMgZWFzeSBhcyBhIHJzeW5jIGZyb20gaW5zaWRl IHRoZSBndWVzdCB0byBhIHJlbW90ZSBtYWNoaW5lLg0KDQpUaGUgYmlnZ2VzdCBwcm9zIGhlcmUg aXMgdGhlIGVhc2Ugb2YgYmVpbmcgYWJsZSB0byBzZXR1cCBhbiBleHBvcnQgZG9tYWluLCBhdHRh Y2gsIGV4cG9ydCB0aGUgVk0sIGRldGFjaCBkb21haW4sIGFuZCB0aGVuIGF0dGFjaCBhbmQgaW1w b3J0IHRvIGEgInJlYWwiIHNldHVwIHdoZW4gdGhlIEFJTyBzdGFydHMgZmVlbGluZyBjcm93ZGVk IGxhdGVyIG9uLiBUaGlua2luZyBhaGVhZCBpcyBuZXZlciBhIGJhZCB0aGluZywgbm8/DQoNCi9L DQo= --_000_5F9E965F5A80BC468BE5F40576769F098C555F4Cexchange21_ Content-Type: text/html; charset="utf-8" Content-ID: <859545D6595E94408D29C283B48B6E99@ad.slu.se> Content-Transfer-Encoding: base64 PGh0bWw+DQo8aGVhZD4NCjxtZXRhIGh0dHAtZXF1aXY9IkNvbnRlbnQtVHlwZSIgY29udGVudD0i dGV4dC9odG1sOyBjaGFyc2V0PXV0Zi04Ij4NCjwvaGVhZD4NCjxib2R5Pg0KPHAgZGlyPSJsdHIi Pjxicj4NCkRlbiAxMiBqdWwgMjAxNCAyMjo0OSBza3JldiBOaWtsYXMgRm9uZGJlcmcgJmx0O25p a2xhc0B2aXJlb25lLmNvbSZndDs6PGJyPg0KJmd0Ozxicj4NCiZndDs8YnI+DQomZ3Q7PGJyPg0K Jmd0OyBPbiAxMiBqdWwgMjAxNCwgYXQgMTY6NTcsICZxdW90O0thcmxpIFNqw7ZiZXJnJnF1b3Q7 ICZsdDtLYXJsaS5Tam9iZXJnQHNsdS5zZSZndDsgd3JvdGU6PGJyPg0KJmd0Ozxicj4NCiZndDsm Z3Q7PGJyPg0KJmd0OyZndDsgRGVuIDEyIGp1bCAyMDE0IDE1OjQ1IHNrcmV2IE5pa2xhcyBGb25k YmVyZyAmbHQ7bmlrbGFzQHZpcmVvbmUuY29tJmd0Ozo8YnI+DQomZ3Q7Jmd0OyAmZ3Q7PGJyPg0K Jmd0OyZndDsgJmd0OyBIaSw8YnI+DQomZ3Q7Jmd0OyAmZ3Q7PGJyPg0KJmd0OyZndDsgJmd0OyBJ 4oCZbSBuZXcgdG8gb1ZpcnQgYnV0IEkgbXVzdCBzYXkgSSBhbSBpbXByZXNzZWQhJm5ic3A7PGJy Pg0KJmd0OyZndDsgJmd0OyBJIGFtIHJ1bm5pbmcgaXQgb24gYSBIUCBETDM4MCB3aXRoIGFuIGV4 dGVybmFsIFNBUyBjaGFzc2kuPGJyPg0KJmd0OyZndDsgJmd0OyBMaW51eCBkaXN0IGlzIENlbnRv cyA2LjUgYW5kIG9WaXJ0IGlzIDMuNCBydW5uaW5nIGFsbC1pbi1vbmUgKGZvciBub3cgdW50aWwg d2UgbmVlZCB0byBoYXZlIGEgc2Vjb25kIGhvc3QpLjxicj4NCiZndDsmZ3Q7ICZndDs8YnI+DQom Z3Q7Jmd0OyAmZ3Q7IE91ciBjb21wYW55ICh3d3cudmlyZW9uZS5jb20pIGRlYWxzIHdpdGggc3lz dGVtIGFyY2hpdGVjdHVyZSBmb3IgbWFueSB0ZWxjbyBhbmQgbWVkaWEgb3BlcmF0b3JzIGFuZCBp cyBub3cgc2V0dGluZyB1cCBhIHNtYWxsIG93biBkYXRhY2VudGVyIGZvciBvdXIgaW50ZXJuYWwg dGVzdHMgYXMgd2VsbCBhcyBvdXIgSVQgaW5mcmFzdHJ1Y3R1cmUuPGJyPg0KJmd0OyZndDsgJmd0 OyBXZSBhcmUgaW4gdGhlIHByb2Nlc3Mgb2YgaW5zdGFsbGluZyBaZW50eWFsIGZvciB0aGUgU01C IHB1cnBvc2VzIG9uIGEgZ3Vlc3QgYW5kIGl0IHdvdWxkIGJlIGdyZWF0IHRvIGhhdmUgdGhhdCBn dWVzdCBhbHNvIHNlcnZpbmcgYSBmcyBwYXRoIGRpcmVjdG9yeSB3aXRoIE5GUyAmIzQzOyBTTUIg KHdoaWNoIGlzIHNlbWkgY3JpcHBsZWQgb24gdGhlIGhvc3QgYWZ0ZXIgb1ZpcnQgaW5zdGFsbGF0 aW9uIHdpdGggdmVyc2lvbiAzIGV0LmMuKS48YnI+DQomZ3Q7Jmd0OyAmZ3Q7PGJyPg0KJmd0OyZn dDsgJmd0OyBEb2VzIGFueW9uZSBoYXZlIGFuIGlkZWEgb2YgaG93IEkgY2FuIHRocm91Z2ggb1Zp cnQgKHNlZW4gc2V2ZXJhbCBzb2x1dGlvbnMgdXNpbmcgdmlyc2ggYW5kIGt2bSkgbGV0dGluZyBt eSBaZW50eWFsIFVidW50dSBndWVzdCBoYXZlIGFjY2VzcyB0byBhIGhvc3QgbW91bnQgcG9pbnQg b3IgaWYgbmVjZXNzYXJ5IChzZWNvbmQgYmVzdCkgYSBzZXBlcmF0ZSBwYXJ0aXRpb24/PGJyPg0K Jmd0OyZndDsgJmd0Ozxicj4NCiZndDsmZ3Q7ICZndDsgQmVzdCByZWdhcmRzPGJyPg0KJmd0OyZn dDsgJmd0OyBOaWtsYXM8YnI+DQomZ3Q7Jmd0OyAmZ3Q7PGJyPg0KJmd0OyZndDs8YnI+DQomZ3Q7 Jmd0OyBXaHkgbm90IGp1c3QgZ2l2ZSB0aGUgZ3Vlc3QgYSB0aGluIHByb3Zpc2lvbiB2aXJ0dWFs IGhhcmQgZHJpdmUgYW5kIGV4cGFuZCBpdCBvbiBhIGRlbWFuZCBiYXNpcz88YnI+DQomZ3Q7Jmd0 Ozxicj4NCiZndDsmZ3Q7IC9LPGJyPg0KJmd0Ozxicj4NCiZndDsgVGhhbmtzIGZvciB0aGUgYWR2 aXNlIGJ1dCB0aGlzIHdvdWxkIG5vdCBzdWl0ZSB1cyBJJ20gYWZyYWlkLiBJdCB3b3VsZCBiZSBk aWZmaWN1bHQgd3J0IGluY3JlbWVudGFsIGJhY2t1cHMgYXMgd2VsbCBhcyBob3N0IG1hY2hpbmUg ZmlsZS1yb3V0aW5lcy4mbmJzcDs8YnI+DQomZ3Q7PGJyPg0KJmd0OzwvcD4NCjxwIGRpcj0ibHRy Ij5XZWxsLCBnb2luZyBieSBPY2NhbSdzIHJhaXpvcjsgdGhlIHNpbXBsZXN0IGFuc3dlciBpcyB1 c3VhbGx5IGNvcnJlY3QuIENhbid0IHJlYWxseSB0ZWxsIHdoYXQgeW91IG1lYW4gYnkgZmlsZS1y b3V0aW5lcyBidXQgYmFja3VwcyB3b3VsZCBiZSB3ZWxsIHNlcnZlZCBieSBzbmFwc2hvdHMgKGNh bid0IGdldCBtb3JlIGluY3JlbWVudGFsIHRoYW4gdGhhdCkgYW5kIGRpc2FzdGVyIHJlY292ZXJ5 IGNvdWxkIGJlIGFzIGVhc3kNCiBhcyBhIHJzeW5jIGZyb20gaW5zaWRlIHRoZSBndWVzdCB0byBh IHJlbW90ZSBtYWNoaW5lLjwvcD4NCjxwIGRpcj0ibHRyIj5UaGUgYmlnZ2VzdCBwcm9zIGhlcmUg aXMgdGhlIGVhc2Ugb2YgYmVpbmcgYWJsZSB0byBzZXR1cCBhbiBleHBvcnQgZG9tYWluLCBhdHRh Y2gsIGV4cG9ydCB0aGUgVk0sIGRldGFjaCBkb21haW4sIGFuZCB0aGVuIGF0dGFjaCBhbmQgaW1w b3J0IHRvIGEgJnF1b3Q7cmVhbCZxdW90OyBzZXR1cCB3aGVuIHRoZSBBSU8gc3RhcnRzIGZlZWxp bmcgY3Jvd2RlZCBsYXRlciBvbi4gVGhpbmtpbmcgYWhlYWQgaXMgbmV2ZXIgYSBiYWQgdGhpbmcs IG5vPzwvcD4NCjxwIGRpcj0ibHRyIj4vSzwvcD4NCjwvYm9keT4NCjwvaHRtbD4NCg== --_000_5F9E965F5A80BC468BE5F40576769F098C555F4Cexchange21_--
--_000_CFE875E3219BBniklasvireonecom_ Content-Type: text/plain; charset="iso-8859-1" Content-Transfer-Encoding: quoted-printable From: Karli Sj=F6berg <Karli.Sjoberg@slu.se<mailto:Karli.Sjoberg@slu.se>> Date: Sunday 13 July 2014 14:51 To: Niklas Fondberg <niklas@vireone.com<mailto:niklas@vireone.com>> Cc: "users@ovirt.org<mailto:users@ovirt.org>" <users@ovirt.org<mailto:users= @ovirt.org>>, Karli Sj=F6berg <Karli.Sjoberg@slu.se<mailto:Karli.Sjoberg@sl= u.se>> Subject: Re: [ovirt-users] fileserver as a guest oVirt Den 12 jul 2014 22:49 skrev Niklas Fondberg <niklas@vireone.com<mailto:nikl= as@vireone.com>>:
On 12 jul 2014, at 16:57, "Karli Sj=F6berg" <Karli.Sjoberg@slu.se<mailto:=
Karli.Sjoberg@slu.se>> wrote:
Den 12 jul 2014 15:45 skrev Niklas Fondberg <niklas@vireone.com<mailto:n=
iklas@vireone.com>>:
Hi,
I'm new to oVirt but I must say I am impressed! I am running it on a HP DL380 with an external SAS chassi. Linux dist is Centos 6.5 and oVirt is 3.4 running all-in-one (for now =
until we need to have a second host).
Our company (www.vireone.com) deals with system architecture for many =
telco and media operators and is now setting up a small own datacenter for = our internal tests as well as our IT infrastructure.
We are in the process of installing Zentyal for the SMB purposes on a = guest and it would be great to have that guest also serving a fs path direc= tory with NFS + SMB (which is semi crippled on the host after oVirt install= ation with version 3 et.c.).
Does anyone have an idea of how I can through oVirt (seen several solu= tions using virsh and kvm) letting my Zentyal Ubuntu guest have access to a= host mount point or if necessary (second best) a seperate partition?
Best regards Niklas
Why not just give the guest a thin provision virtual hard drive and expa= nd it on a demand basis?
/K
Thanks for the advise but this would not suite us I'm afraid. It would be= difficult wrt incremental backups as well as host machine file-routines.
Well, going by Occam's raizor; the simplest answer is usually correct. Can'= t really tell what you mean by file-routines but backups would be well serv= ed by snapshots (can't get more incremental than that) and disaster recover= y could be as easy as a rsync from inside the guest to a remote machine. The biggest pros here is the ease of being able to setup an export domain, = attach, export the VM, detach domain, and then attach and import to a "real= " setup when the AIO starts feeling crowded later on. Thinking ahead is nev= er a bad thing, no? /K Thanks for your suggestions! The thing also is that the performance will be very bad if we have the 25TB= SAS array shared for our purposes (lots of media streaming) using a virtua= l disk. What I am after (after more reading) is support for virtio-9p-pci (http://w= ww.linux-kvm.org/page/9p_virtio) using oVirt. Alternative is the Direct LUN= hook (http://www.ovirt.org/VDSM-Hooks/directlun, if I can figure out how t= o work with hooks...) Any chance anybody has an answer for these questions? --_000_CFE875E3219BBniklasvireonecom_ Content-Type: text/html; charset="iso-8859-1" Content-ID: <5A76E9D251C9A743BC133AC39E2E7684@eurprd04.prod.outlook.com> Content-Transfer-Encoding: quoted-printable <html> <head> <meta http-equiv=3D"Content-Type" content=3D"text/html; charset=3Diso-8859-= 1"> </head> <body style=3D"word-wrap: break-word; -webkit-nbsp-mode: space; -webkit-lin= e-break: after-white-space; color: rgb(0, 0, 0); font-size: 14px; font-fami= ly: Calibri, sans-serif;"> <div><br> </div> <div><br> </div> <span id=3D"OLK_SRC_BODY_SECTION"> <div style=3D"font-family:Calibri; font-size:11pt; text-align:left; color:b= lack; BORDER-BOTTOM: medium none; BORDER-LEFT: medium none; PADDING-BOTTOM:= 0in; PADDING-LEFT: 0in; PADDING-RIGHT: 0in; BORDER-TOP: #b5c4df 1pt solid;= BORDER-RIGHT: medium none; PADDING-TOP: 3pt"> <span style=3D"font-weight:bold">From: </span>Karli Sj=F6berg <<a href= =3D"mailto:Karli.Sjoberg@slu.se">Karli.Sjoberg@slu.se</a>><br> <span style=3D"font-weight:bold">Date: </span>Sunday 13 July 2014 14:51<br> <span style=3D"font-weight:bold">To: </span>Niklas Fondberg <<a href=3D"= mailto:niklas@vireone.com">niklas@vireone.com</a>><br> <span style=3D"font-weight:bold">Cc: </span>"<a href=3D"mailto:users@o= virt.org">users@ovirt.org</a>" <<a href=3D"mailto:users@ovirt.org">= users@ovirt.org</a>>, Karli Sj=F6berg <<a href=3D"mailto:Karli.Sjober= g@slu.se">Karli.Sjoberg@slu.se</a>><br> <span style=3D"font-weight:bold">Subject: </span>Re: [ovirt-users] fileserv= er as a guest oVirt<br> </div> <div><br> </div> <blockquote id=3D"MAC_OUTLOOK_ATTRIBUTION_BLOCKQUOTE" style=3D"BORDER-LEFT:= #b5c4df 5 solid; PADDING:0 0 0 5; MARGIN:0 0 0 5;"> <div> <div> <p dir=3D"ltr"><br> Den 12 jul 2014 22:49 skrev Niklas Fondberg <<a href=3D"mailto:niklas@vi= reone.com">niklas@vireone.com</a>>:<br> ><br> ><br> ><br> > On 12 jul 2014, at 16:57, "Karli Sj=F6berg" <<a href=3D"m= ailto:Karli.Sjoberg@slu.se">Karli.Sjoberg@slu.se</a>> wrote:<br> ><br> >><br> >> Den 12 jul 2014 15:45 skrev Niklas Fondberg <<a href=3D"mailto:= niklas@vireone.com">niklas@vireone.com</a>>:<br> >> ><br> >> > Hi,<br> >> ><br> >> > I’m new to oVirt but I must say I am impressed! <b= r> >> > I am running it on a HP DL380 with an external SAS chassi.<br=
>> > Linux dist is Centos 6.5 and oVirt is 3.4 running all-in-one = (for now until we need to have a second host).<br> >> ><br> >> > Our company (www.vireone.com) deals with system architecture = for many telco and media operators and is now setting up a small own datace= nter for our internal tests as well as our IT infrastructure.<br> >> > We are in the process of installing Zentyal for the SMB purpo= ses on a guest and it would be great to have that guest also serving a fs p= ath directory with NFS + SMB (which is semi crippled on the host after = oVirt installation with version 3 et.c.).<br> >> ><br> >> > Does anyone have an idea of how I can through oVirt (seen sev= eral solutions using virsh and kvm) letting my Zentyal Ubuntu guest have ac= cess to a host mount point or if necessary (second best) a seperate partiti= on?<br> >> ><br> >> > Best regards<br> >> > Niklas<br> >> ><br> >><br> >> Why not just give the guest a thin provision virtual hard drive an= d expand it on a demand basis?<br> >><br> >> /K<br> ><br> > Thanks for the advise but this would not suite us I'm afraid. It would= be difficult wrt incremental backups as well as host machine file-routines= . <br> ><br> ></p> <p dir=3D"ltr">Well, going by Occam's raizor; the simplest answer is usuall= y correct. Can't really tell what you mean by file-routines but backups wou= ld be well served by snapshots (can't get more incremental than that) and d= isaster recovery could be as easy as a rsync from inside the guest to a remote machine.</p> <p dir=3D"ltr">The biggest pros here is the ease of being able to setup an = export domain, attach, export the VM, detach domain, and then attach and im= port to a "real" setup when the AIO starts feeling crowded later = on. Thinking ahead is never a bad thing, no?</p> <p dir=3D"ltr">/K</p> </div> </div> </blockquote> </span> <div>Thanks for your suggestions!</div> <div>The thing also is that the performance will be very bad if we have the= 25TB SAS array shared for our purposes (lots of media streaming) using a v= irtual disk. </div> <div>What I am after (after more reading) is support for <span style= =3D"background-color: rgb(255, 255, 255); color: rgb(67, 70, 86);">virtio-9= p-pci (</span><a href=3D"http://www.linux-kvm.org/page/9p_virtio">http://ww= w.linux-kvm.org/page/9p_virtio</a>) using oVirt. Alternative is the Direct LUN hook (<a href=3D"http://www.ovirt.org/VDSM-H= ooks/directlun">http://www.ovirt.org/VDSM-Hooks/directlun</a>, if I ca= n figure out how to work with hooks…)</div> <div>Any chance anybody has an answer for these questions?</div> <div><br> </div> </body> </html> --_000_CFE875E3219BBniklasvireonecom_--
=20 From: Karli Sj=C3=B6berg <Karli.Sjoberg@slu.se<mailto:Karli.Sjoberg@slu.s= e>> Date: Sunday 13 July 2014 14:51 To: Niklas Fondberg <niklas@vireone.com<mailto:niklas@vireone.com>> Cc: "users@ovirt.org<mailto:users@ovirt.org>" <users@ovirt.org<mailto:use= rs@ovirt.org>>, Karli Sj=C3=B6berg <Karli.Sjoberg@slu.se<mailto:Karli.Sjobe= rg@slu.se>> Subject: Re: [ovirt-users] fileserver as a guest oVirt =20 =20 Den 12 jul 2014 22:49 skrev Niklas Fondberg <niklas@vireone.com<mailto:ni= klas@vireone.com>>:
On 12 jul 2014, at 16:57, "Karli Sj=C3=B6berg" <Karli.Sjoberg@slu.se<ma=
ilto:Karli.Sjoberg@slu.se>> wrote:
Den 12 jul 2014 15:45 skrev Niklas Fondberg <niklas@vireone.com<mailto=
:niklas@vireone.com>>:
Hi,
I'm new to oVirt but I must say I am impressed! I am running it on a HP DL380 with an external SAS chassi. Linux dist is Centos 6.5 and oVirt is 3.4 running all-in-one (for no=
w until we need to have a second host).
Our company (www.vireone.com) deals with system architecture for man=
y telco and media operators and is now setting up a small own datacenter fo= r our internal tests as well as our IT infrastructure.
We are in the process of installing Zentyal for the SMB purposes on = a guest and it would be great to have that guest also serving a fs path dir= ectory with NFS + SMB (which is semi crippled on the host after oVirt insta= llation with version 3 et.c.).
Does anyone have an idea of how I can through oVirt (seen several so= lutions using virsh and kvm) letting my Zentyal Ubuntu guest have access to= a host mount point or if necessary (second best) a seperate partition?
Best regards Niklas
Why not just give the guest a thin provision virtual hard drive and ex=
--=-QO4YC7yJVpEFWdlCUWRr Content-Type: text/plain; charset="UTF-8" Content-Transfer-Encoding: quoted-printable Hello, just add my 2ct: I did a lot of bench marking for our SAN (FC LUN's). I also need file servers for our SMB Clients. I recommend using Direct Attached LUNs for your purpose and attach them to the VMs as VirtIO-SCSI disks. You can even added them as shareable to oVirt if you deploy some kind of SAN file system (we use Quantum's StorNext). Bottom line, the implementation of VirtIO-SCSI is so well done and support in oVirt is great. I cound not see any bottlenecks in the visualization. For the foreseeable future I will not deploy bare metal file Servers again. HTH, On So, 2014-07-13 at 15:47 +0000, Niklas Fondberg wrote: pand it on a demand basis?
/K
Thanks for the advise but this would not suite us I'm afraid. It would = be difficult wrt incremental backups as well as host machine file-routines.
=20 Well, going by Occam's raizor; the simplest answer is usually correct. Ca= n't really tell what you mean by file-routines but backups would be well se= rved by snapshots (can't get more incremental than that) and disaster recov= ery could be as easy as a rsync from inside the guest to a remote machine. =20 The biggest pros here is the ease of being able to setup an export domain= , attach, export the VM, detach domain, and then attach and import to a "re= al" setup when the AIO starts feeling crowded later on. Thinking ahead is n= ever a bad thing, no? =20 /K =20 Thanks for your suggestions! The thing also is that the performance will be very bad if we have the 25= TB SAS array shared for our purposes (lots of media streaming) using a virt= ual disk. What I am after (after more reading) is support for virtio-9p-pci (http:/= /www.linux-kvm.org/page/9p_virtio) using oVirt. Alternative is the Direct L= UN hook (http://www.ovirt.org/VDSM-Hooks/directlun, if I can figure out how= to work with hooks...) Any chance anybody has an answer for these questions? =20 _______________________________________________ Users mailing list Users@ovirt.org http://lists.ovirt.org/mailman/listinfo/users
--=20 Daniel Helgenberger=20 m box bewegtbild GmbH=20 P: +49/30/2408781-22 F: +49/30/2408781-10 ACKERSTR. 19=20 D-10115 BERLIN=20 www.m-box.de www.monkeymen.tv=20 Gesch=C3=A4ftsf=C3=BChrer: Martin Retschitzegger / Michaela G=C3=B6llner Handeslregister: Amtsgericht Charlottenburg / HRB 112767=20 --=-QO4YC7yJVpEFWdlCUWRr Content-Type: application/x-pkcs7-signature; name="smime.p7s" Content-Disposition: attachment; filename="smime.p7s" Content-Transfer-Encoding: base64 MIAGCSqGSIb3DQEHAqCAMIACAQExCzAJBgUrDgMCGgUAMIAGCSqGSIb3DQEHAQAAoIINtjCCBBYw ggL+oAMCAQICCwQAAAAAAS9O4S9SMA0GCSqGSIb3DQEBBQUAMFcxCzAJBgNVBAYTAkJFMRkwFwYD VQQKExBHbG9iYWxTaWduIG52LXNhMRAwDgYDVQQLEwdSb290IENBMRswGQYDVQQDExJHbG9iYWxT aWduIFJvb3QgQ0EwHhcNMTEwNDEzMTAwMDAwWhcNMTkwNDEzMTAwMDAwWjBUMQswCQYDVQQGEwJC RTEZMBcGA1UEChMQR2xvYmFsU2lnbiBudi1zYTEqMCgGA1UEAxMhR2xvYmFsU2lnbiBQZXJzb25h bFNpZ24gMiBDQSAtIEcyMIIBIjANBgkqhkiG9w0BAQEFAAOCAQ8AMIIBCgKCAQEAwWtB+TXs+BJ9 3SJRaV+3uRNGJ3cUO+MTgW8+5HQXfgy19CzkDI1T1NwwICi/bo4R/mYR5FEWx91//eE0ElC/89iY 7GkL0tDasmVx4TOXnrqrsziUcxEPPqHRE8x4NhtBK7+8o0nsMIJMA1gyZ2FA5To2Ew1BBuvovvDJ +Nua3qOCNBNu+8A+eNpJlVnlu/qB7+XWaPXtUMlsIikxD+gREFVUgYE4VzBuLa2kkg0VLd09XkE2 ceRDm6YgRATuDk6ogUyX4OLxCGIJF8yi6Z37M0wemDA6Uff0EuqdwDQd5HwG/rernUjt1grLdAxq 8BwywRRg0eFHmE+ShhpyO3Fi+wIDAQABo4HlMIHiMA4GA1UdDwEB/wQEAwIBBjASBgNVHRMBAf8E CDAGAQH/AgEAMB0GA1UdDgQWBBQ/FdJtfC/nMZ5DCgaolGwsO8XuZTBHBgNVHSAEQDA+MDwGBFUd IAAwNDAyBggrBgEFBQcCARYmaHR0cHM6Ly93d3cuZ2xvYmFsc2lnbi5jb20vcmVwb3NpdG9yeS8w MwYDVR0fBCwwKjAooCagJIYiaHR0cDovL2NybC5nbG9iYWxzaWduLm5ldC9yb290LmNybDAfBgNV HSMEGDAWgBRge2YaRQ2XyolQL30EzTSo//z9SzANBgkqhkiG9w0BAQUFAAOCAQEAQ3N5zKTMSTED HGFAgd/gu91Kb8AxPHgjq+7dhf7mkCinMqqrLai2XOrz8CP63BPaAx7oGOUBI0MyASBGk5zej9L3 oHtiF2BL01m1sBnT8rQxT2CJd/+jqjUl0p2ew8T3HSyatrsooGvDwf00yCB2JHTNvtQxNO8t6x/+ 048A1Q+0i7uf0nTnyrJLjD04zhL89ytetZspltOpJVYbmwiFjq6PxsdUNthUDme/9pOLmKDnQU0p W/JEwLs2TYCBNKwdgSGAk8/z+s2SCltKIG0Uh5U6t6j7JPuwNP/znImwMrlHDJ1YpW0rkF2PGraV CgDBf9dOB+IIpnwHfIi+LD+eITCCBMowggOyoAMCAQICEQCWaWbA3qWpL+Qmn6I16DynMA0GCSqG SIb3DQEBBQUAMFQxCzAJBgNVBAYTAkJFMRkwFwYDVQQKExBHbG9iYWxTaWduIG52LXNhMSowKAYD VQQDEyFHbG9iYWxTaWduIFBlcnNvbmFsU2lnbiAyIENBIC0gRzIwHhcNMTMwODI3MTY1NzU4WhcN MTYwODI3MTY1NzU4WjBYMQswCQYDVQQGEwJERTEcMBoGA1UEAxMTRGFuaWVsIEhlbGdlbmJlcmdl cjErMCkGCSqGSIb3DQEJARYcZGFuaWVsLmhlbGdlbmJlcmdlckBtLWJveC5kZTCCASIwDQYJKoZI hvcNAQEBBQADggEPADCCAQoCggEBAM4BQ5vPknk1OGLd1qKSUIKmQLrjccjJcYj7qtAtA+fNYKF8 9p1VY4UwiFcF9jKlmA9Q8o8tYSx16LYYFoGWokNRAeKFXZiBZiHyI0ekpEfxo8N5cTMCcxKcSYWV 8sqzmBPCoMNpmiVoC8ec8Nv5SqXH34VVtDmNLfiVlsTyomBXAJkJ2/n5XqJzPLFGWWREtPLkVVS+ u426vt/hNsQi5akNoidYeXo98JcrmeApFJ3zB2KxvMziHx8LD4q1gAl9NumtX5YLbCpdWL9AkWdX Oaro3D9zj6Q6LyGwa/UQUrZdg3BXc07hjHZn6d9vet1SzpbyqQpTzM63yXiX1meEMlMCAwEAAaOC AZEwggGNMA4GA1UdDwEB/wQEAwIFoDBMBgNVHSAERTBDMEEGCSsGAQQBoDIBKDA0MDIGCCsGAQUF BwIBFiZodHRwczovL3d3dy5nbG9iYWxzaWduLmNvbS9yZXBvc2l0b3J5LzAnBgNVHREEIDAegRxk YW5pZWwuaGVsZ2VuYmVyZ2VyQG0tYm94LmRlMAkGA1UdEwQCMAAwHQYDVR0lBBYwFAYIKwYBBQUH AwIGCCsGAQUFBwMEMEMGA1UdHwQ8MDowOKA2oDSGMmh0dHA6Ly9jcmwuZ2xvYmFsc2lnbi5jb20v Z3MvZ3NwZXJzb25hbHNpZ24yZzIuY3JsMFUGCCsGAQUFBwEBBEkwRzBFBggrBgEFBQcwAoY5aHR0 cDovL3NlY3VyZS5nbG9iYWxzaWduLmNvbS9jYWNlcnQvZ3NwZXJzb25hbHNpZ24yZzIuY3J0MB0G A1UdDgQWBBS8NFA/upd+Wipw2nj8RD/Ct+R2GTAfBgNVHSMEGDAWgBQ/FdJtfC/nMZ5DCgaolGws O8XuZTANBgkqhkiG9w0BAQUFAAOCAQEAXVTpu4fhOLETAW0zdbQiIwBIMZgeVNJnWV3GsMxByycU 63P+WBQTBl9qj47vHLmVdeF7MzH0QSXZSc9Tnfr6CYIImpyIZxRAGpAsWmtZf3JieRA0+j4GQJF2 zAea1NXYXoG9+ZSSZHBSxKUdrRdVdE320nuVGTT2HjEI2LEYbOvaXyi6HhpuHUiyu4LD0+RIT3fi T8jUiKKLTsApTD+Ak8SLF0IESOSA6htirv69mDDC7Klg9dT7QBPO7dpoKIUOldV3VhahndVfsDff KD7pkUUvG5XftYEQOxlWDJzuTBeqf/4hxXMtzFU9OaI6oKJjLfr6B+XBc6xwOtc/NMWmejCCBMow ggOyoAMCAQICEQCWaWbA3qWpL+Qmn6I16DynMA0GCSqGSIb3DQEBBQUAMFQxCzAJBgNVBAYTAkJF MRkwFwYDVQQKExBHbG9iYWxTaWduIG52LXNhMSowKAYDVQQDEyFHbG9iYWxTaWduIFBlcnNvbmFs U2lnbiAyIENBIC0gRzIwHhcNMTMwODI3MTY1NzU4WhcNMTYwODI3MTY1NzU4WjBYMQswCQYDVQQG EwJERTEcMBoGA1UEAxMTRGFuaWVsIEhlbGdlbmJlcmdlcjErMCkGCSqGSIb3DQEJARYcZGFuaWVs LmhlbGdlbmJlcmdlckBtLWJveC5kZTCCASIwDQYJKoZIhvcNAQEBBQADggEPADCCAQoCggEBAM4B Q5vPknk1OGLd1qKSUIKmQLrjccjJcYj7qtAtA+fNYKF89p1VY4UwiFcF9jKlmA9Q8o8tYSx16LYY FoGWokNRAeKFXZiBZiHyI0ekpEfxo8N5cTMCcxKcSYWV8sqzmBPCoMNpmiVoC8ec8Nv5SqXH34VV tDmNLfiVlsTyomBXAJkJ2/n5XqJzPLFGWWREtPLkVVS+u426vt/hNsQi5akNoidYeXo98JcrmeAp FJ3zB2KxvMziHx8LD4q1gAl9NumtX5YLbCpdWL9AkWdXOaro3D9zj6Q6LyGwa/UQUrZdg3BXc07h jHZn6d9vet1SzpbyqQpTzM63yXiX1meEMlMCAwEAAaOCAZEwggGNMA4GA1UdDwEB/wQEAwIFoDBM BgNVHSAERTBDMEEGCSsGAQQBoDIBKDA0MDIGCCsGAQUFBwIBFiZodHRwczovL3d3dy5nbG9iYWxz aWduLmNvbS9yZXBvc2l0b3J5LzAnBgNVHREEIDAegRxkYW5pZWwuaGVsZ2VuYmVyZ2VyQG0tYm94 LmRlMAkGA1UdEwQCMAAwHQYDVR0lBBYwFAYIKwYBBQUHAwIGCCsGAQUFBwMEMEMGA1UdHwQ8MDow OKA2oDSGMmh0dHA6Ly9jcmwuZ2xvYmFsc2lnbi5jb20vZ3MvZ3NwZXJzb25hbHNpZ24yZzIuY3Js MFUGCCsGAQUFBwEBBEkwRzBFBggrBgEFBQcwAoY5aHR0cDovL3NlY3VyZS5nbG9iYWxzaWduLmNv bS9jYWNlcnQvZ3NwZXJzb25hbHNpZ24yZzIuY3J0MB0GA1UdDgQWBBS8NFA/upd+Wipw2nj8RD/C t+R2GTAfBgNVHSMEGDAWgBQ/FdJtfC/nMZ5DCgaolGwsO8XuZTANBgkqhkiG9w0BAQUFAAOCAQEA XVTpu4fhOLETAW0zdbQiIwBIMZgeVNJnWV3GsMxByycU63P+WBQTBl9qj47vHLmVdeF7MzH0QSXZ Sc9Tnfr6CYIImpyIZxRAGpAsWmtZf3JieRA0+j4GQJF2zAea1NXYXoG9+ZSSZHBSxKUdrRdVdE32 0nuVGTT2HjEI2LEYbOvaXyi6HhpuHUiyu4LD0+RIT3fiT8jUiKKLTsApTD+Ak8SLF0IESOSA6hti rv69mDDC7Klg9dT7QBPO7dpoKIUOldV3VhahndVfsDffKD7pkUUvG5XftYEQOxlWDJzuTBeqf/4h xXMtzFU9OaI6oKJjLfr6B+XBc6xwOtc/NMWmejGCAucwggLjAgEBMGkwVDELMAkGA1UEBhMCQkUx GTAXBgNVBAoTEEdsb2JhbFNpZ24gbnYtc2ExKjAoBgNVBAMTIUdsb2JhbFNpZ24gUGVyc29uYWxT aWduIDIgQ0EgLSBHMgIRAJZpZsDepakv5CafojXoPKcwCQYFKw4DAhoFAKCCAVMwGAYJKoZIhvcN AQkDMQsGCSqGSIb3DQEHATAcBgkqhkiG9w0BCQUxDxcNMTQwNzE0MDY1NzQwWjAjBgkqhkiG9w0B CQQxFgQUsr0rsg0NKIeUjm/xPdT43nUTAH8weAYJKwYBBAGCNxAEMWswaTBUMQswCQYDVQQGEwJC RTEZMBcGA1UEChMQR2xvYmFsU2lnbiBudi1zYTEqMCgGA1UEAxMhR2xvYmFsU2lnbiBQZXJzb25h bFNpZ24gMiBDQSAtIEcyAhEAlmlmwN6lqS/kJp+iNeg8pzB6BgsqhkiG9w0BCRACCzFroGkwVDEL MAkGA1UEBhMCQkUxGTAXBgNVBAoTEEdsb2JhbFNpZ24gbnYtc2ExKjAoBgNVBAMTIUdsb2JhbFNp Z24gUGVyc29uYWxTaWduIDIgQ0EgLSBHMgIRAJZpZsDepakv5CafojXoPKcwDQYJKoZIhvcNAQEB BQAEggEAl48Pw50qzXdvYsn0TA69RRFfy1+3JL3tzPDorgUERs9/P1prbdZih7RYe+bim/gH78BT a1IH+HAGDjPonkxGM5re23vIkAiSYswCLkB/gYWXZwOMGZpXq6POhCFuWKyP6cal89ghCNdOfI2z EcUhazJWQJy/3TPHzdRIV+taTDXOue3KthjURciacxBOGubyTn5e9xP9rgmEjgTmHdAnG0u0K9R5 eB4U31hh7sd9jQ4jN19kxysr4n+YT9E7bwr5zQHyUAFDWr3TEyJUEiHXftOiGWsc1qnszf2QSzWr bENfKZLQHHQrocnvFf2tOQ/sH0GmjLGGUBmgnhN3EcgwuQAAAAAAAA== --=-QO4YC7yJVpEFWdlCUWRr--
Thanks, after reading it makes sense. I suppose I need to drop my hope of having sharing possibility with the host. I have two questions that you might be able to answer: 1. Does direct LUN support partitions or only whole devices 2. Do you know of any open source way of making them shareable to Ovirt? My setup is simple: - HP DL380 with dual Xeon and lots of RAM. Boots from seperate disk (USB) - MSA60 with p411 attached to the HP DL380 When we grow oVirt we will grow with DL360¹s and use all shared storage from the guest fileserver on the first host. On 14/07/14 08:57, "Daniel Helgenberger" <daniel.helgenberger@m-box.de> wrote:
Hello,
just add my 2ct: I did a lot of bench marking for our SAN (FC LUN's). I also need file servers for our SMB Clients.
I recommend using Direct Attached LUNs for your purpose and attach them to the VMs as VirtIO-SCSI disks. You can even added them as shareable to oVirt if you deploy some kind of SAN file system (we use Quantum's StorNext).
Bottom line, the implementation of VirtIO-SCSI is so well done and support in oVirt is great. I cound not see any bottlenecks in the visualization. For the foreseeable future I will not deploy bare metal file Servers again.
HTH,
On So, 2014-07-13 at 15:47 +0000, Niklas Fondberg wrote:
From: Karli Sjöberg <Karli.Sjoberg@slu.se<mailto:Karli.Sjoberg@slu.se>> Date: Sunday 13 July 2014 14:51 To: Niklas Fondberg <niklas@vireone.com<mailto:niklas@vireone.com>> Cc: "users@ovirt.org<mailto:users@ovirt.org>" <users@ovirt.org<mailto:users@ovirt.org>>, Karli Sjöberg <Karli.Sjoberg@slu.se<mailto:Karli.Sjoberg@slu.se>> Subject: Re: [ovirt-users] fileserver as a guest oVirt
Den 12 jul 2014 22:49 skrev Niklas Fondberg <niklas@vireone.com<mailto:niklas@vireone.com>>:
On 12 jul 2014, at 16:57, "Karli Sjöberg"
<Karli.Sjoberg@slu.se<mailto:Karli.Sjoberg@slu.se>> wrote:
Den 12 jul 2014 15:45 skrev Niklas Fondberg
Hi,
I'm new to oVirt but I must say I am impressed! I am running it on a HP DL380 with an external SAS chassi. Linux dist is Centos 6.5 and oVirt is 3.4 running all-in-one (for
now until we need to have a second host).
Our company (www.vireone.com) deals with system architecture for
many telco and media operators and is now setting up a small own datacenter for our internal tests as well as our IT infrastructure.
We are in the process of installing Zentyal for the SMB purposes on a guest and it would be great to have that guest also serving a fs
Does anyone have an idea of how I can through oVirt (seen several
solutions using virsh and kvm) letting my Zentyal Ubuntu guest have access to a host mount point or if necessary (second best) a seperate
<niklas@vireone.com<mailto:niklas@vireone.com>>: path directory with NFS + SMB (which is semi crippled on the host after oVirt installation with version 3 et.c.). partition?
Best regards Niklas
Why not just give the guest a thin provision virtual hard drive and expand it on a demand basis?
/K
Thanks for the advise but this would not suite us I'm afraid. It would be difficult wrt incremental backups as well as host machine file-routines.
Well, going by Occam's raizor; the simplest answer is usually correct. Can't really tell what you mean by file-routines but backups would be well served by snapshots (can't get more incremental than that) and disaster recovery could be as easy as a rsync from inside the guest to a remote machine.
The biggest pros here is the ease of being able to setup an export domain, attach, export the VM, detach domain, and then attach and import to a "real" setup when the AIO starts feeling crowded later on. Thinking ahead is never a bad thing, no?
/K
Thanks for your suggestions! The thing also is that the performance will be very bad if we have the 25TB SAS array shared for our purposes (lots of media streaming) using a virtual disk. What I am after (after more reading) is support for virtio-9p-pci (http://www.linux-kvm.org/page/9p_virtio) using oVirt. Alternative is the Direct LUN hook (http://www.ovirt.org/VDSM-Hooks/directlun, if I can figure out how to work with hooks...) Any chance anybody has an answer for these questions?
_______________________________________________ Users mailing list Users@ovirt.org http://lists.ovirt.org/mailman/listinfo/users
--
Daniel Helgenberger m box bewegtbild GmbH
P: +49/30/2408781-22 F: +49/30/2408781-10
ACKERSTR. 19 D-10115 BERLIN
www.m-box.de www.monkeymen.tv
Geschäftsführer: Martin Retschitzegger / Michaela Göllner Handeslregister: Amtsgericht Charlottenburg / HRB 112767
--=-aZsrVzyAkVB3GVWiu1xW Content-Type: text/plain; charset="UTF-8" Content-Transfer-Encoding: quoted-printable Hello, for some reason the message was flagged spam; so there was a deplay. oVirt supports direct LUNs. These LUNs are often already partitions of some RAID enclosure. AFAIK the MSA60 is a JBOD. You can use the p411 controller to create your partitions / LUNs. The Virtio-SCSI paravirt driver supports a wide range of (=3Dmost) SCSI commands. This way clients can access them as 'real' SCSI devices. If you did partition the LUN with parted, then the client(s) will see these partitions also, along with the file system on it.=20 As you might know, you cannot have 'normal' file systems mounted rw on several machines at one, you need a cluster file system; see examples in [1]; there are several open source FS's around. Also, nothing stops you from mounting a file system read only on several hosts. One note on the subject, though: I consider shared disk file systems as 'old' approach. I support them in our setup because of historical reasons.=20 Today in a new deployment I would tend to use more 'modern' scale out file systems like GlusterFS (support in oVirt is quite well) or Ceph/Rados as object store file system. Using the native clients, you basically have a shared disk file system with less bottlenecks (MDC's in shared disk fs). Also, both examples have APIs - an application using this can greatly benefit in performance. Again, I consider using APIs for file storage the approach of the future. If you need to attach NFS / CIFS clients, you can always reshare these file systems and use (p)NFS or CTDB if you want to cluster this. Note, with one host and one JBOD this makes little sense to me. [1] http://en.wikipedia.org/wiki/Clustered_file_system [2] https://ctdb.samba.org/ On Mo, 2014-07-14 at 08:10 +0000, Niklas Fondberg wrote:
Thanks, after reading it makes sense. I suppose I need to drop my hope of having sharing possibility with the host. I have two questions that you might be able to answer: 1. Does direct LUN support partitions or only whole devices 2. Do you know of any open source way of making them shareable to Ovirt? =20 My setup is simple: - HP DL380 with dual Xeon and lots of RAM. Boots from seperate disk (USB) - MSA60 with p411 attached to the HP DL380 =20 When we grow oVirt we will grow with DL360=C2=B9s and use all shared stor= age from the guest fileserver on the first host. =20 =20 =20 On 14/07/14 08:57, "Daniel Helgenberger" <daniel.helgenberger@m-box.de> wrote: =20
Hello,
just add my 2ct: I did a lot of bench marking for our SAN (FC LUN's). I also need file servers for our SMB Clients.
I recommend using Direct Attached LUNs for your purpose and attach them to the VMs as VirtIO-SCSI disks. You can even added them as shareable to oVirt if you deploy some kind of SAN file system (we use Quantum's StorNext).
Bottom line, the implementation of VirtIO-SCSI is so well done and support in oVirt is great. I cound not see any bottlenecks in the visualization. For the foreseeable future I will not deploy bare metal file Servers again.
HTH,
On So, 2014-07-13 at 15:47 +0000, Niklas Fondberg wrote:
=20 From: Karli Sj=C3=B6berg <Karli.Sjoberg@slu.se<mailto:Karli.Sjoberg@sl= u.se>> Date: Sunday 13 July 2014 14:51 To: Niklas Fondberg <niklas@vireone.com<mailto:niklas@vireone.com>> Cc: "users@ovirt.org<mailto:users@ovirt.org>" <users@ovirt.org<mailto:users@ovirt.org>>, Karli Sj=C3=B6berg <Karli.Sjoberg@slu.se<mailto:Karli.Sjoberg@slu.se>> Subject: Re: [ovirt-users] fileserver as a guest oVirt =20 =20 Den 12 jul 2014 22:49 skrev Niklas Fondberg <niklas@vireone.com<mailto:niklas@vireone.com>>:
On 12 jul 2014, at 16:57, "Karli Sj=C3=B6berg"
<Karli.Sjoberg@slu.se<mailto:Karli.Sjoberg@slu.se>> wrote:
Den 12 jul 2014 15:45 skrev Niklas Fondberg
Hi,
I'm new to oVirt but I must say I am impressed! I am running it on a HP DL380 with an external SAS chassi. Linux dist is Centos 6.5 and oVirt is 3.4 running all-in-one (for
now until we need to have a second host).
Our company (www.vireone.com) deals with system architecture for
many telco and media operators and is now setting up a small own datacenter for our internal tests as well as our IT infrastructure.
We are in the process of installing Zentyal for the SMB purposes on a guest and it would be great to have that guest also serving a fs
Does anyone have an idea of how I can through oVirt (seen several
solutions using virsh and kvm) letting my Zentyal Ubuntu guest have access to a host mount point or if necessary (second best) a seperate
<niklas@vireone.com<mailto:niklas@vireone.com>>: path directory with NFS + SMB (which is semi crippled on the host after oVirt installation with version 3 et.c.). partition?
Best regards Niklas
Why not just give the guest a thin provision virtual hard drive and expand it on a demand basis?
/K
Thanks for the advise but this would not suite us I'm afraid. It would be difficult wrt incremental backups as well as host machine file-routines.
=20 Well, going by Occam's raizor; the simplest answer is usually correct. Can't really tell what you mean by file-routines but backups would be well served by snapshots (can't get more incremental than that) and disaster recovery could be as easy as a rsync from inside the guest to = a remote machine. =20 The biggest pros here is the ease of being able to setup an export domain, attach, export the VM, detach domain, and then attach and impor= t to a "real" setup when the AIO starts feeling crowded later on. Thinkin= g ahead is never a bad thing, no? =20 /K =20 Thanks for your suggestions! The thing also is that the performance will be very bad if we have the 25TB SAS array shared for our purposes (lots of media streaming) using = a virtual disk. What I am after (after more reading) is support for virtio-9p-pci (http://www.linux-kvm.org/page/9p_virtio) using oVirt. Alternative is the Direct LUN hook (http://www.ovirt.org/VDSM-Hooks/directlun, if I ca= n figure out how to work with hooks...) Any chance anybody has an answer for these questions? =20 _______________________________________________ Users mailing list Users@ovirt.org http://lists.ovirt.org/mailman/listinfo/users
--=20
Daniel Helgenberger m box bewegtbild GmbH
P: +49/30/2408781-22 F: +49/30/2408781-10
ACKERSTR. 19=20 D-10115 BERLIN=20
www.m-box.de www.monkeymen.tv
Gesch=C3=A4ftsf=C3=BChrer: Martin Retschitzegger / Michaela G=C3=B6llner Handeslregister: Amtsgericht Charlottenburg / HRB 112767=20 =20
--=20 Daniel Helgenberger=20 m box bewegtbild GmbH=20 P: +49/30/2408781-22 F: +49/30/2408781-10 ACKERSTR. 19=20 D-10115 BERLIN=20 www.m-box.de www.monkeymen.tv=20 Gesch=C3=A4ftsf=C3=BChrer: Martin Retschitzegger / Michaela G=C3=B6llner Handeslregister: Amtsgericht Charlottenburg / HRB 112767=20 --=-aZsrVzyAkVB3GVWiu1xW Content-Type: application/x-pkcs7-signature; name="smime.p7s" Content-Disposition: attachment; filename="smime.p7s" Content-Transfer-Encoding: base64 MIAGCSqGSIb3DQEHAqCAMIACAQExCzAJBgUrDgMCGgUAMIAGCSqGSIb3DQEHAQAAoIINtjCCBBYw ggL+oAMCAQICCwQAAAAAAS9O4S9SMA0GCSqGSIb3DQEBBQUAMFcxCzAJBgNVBAYTAkJFMRkwFwYD VQQKExBHbG9iYWxTaWduIG52LXNhMRAwDgYDVQQLEwdSb290IENBMRswGQYDVQQDExJHbG9iYWxT aWduIFJvb3QgQ0EwHhcNMTEwNDEzMTAwMDAwWhcNMTkwNDEzMTAwMDAwWjBUMQswCQYDVQQGEwJC RTEZMBcGA1UEChMQR2xvYmFsU2lnbiBudi1zYTEqMCgGA1UEAxMhR2xvYmFsU2lnbiBQZXJzb25h bFNpZ24gMiBDQSAtIEcyMIIBIjANBgkqhkiG9w0BAQEFAAOCAQ8AMIIBCgKCAQEAwWtB+TXs+BJ9 3SJRaV+3uRNGJ3cUO+MTgW8+5HQXfgy19CzkDI1T1NwwICi/bo4R/mYR5FEWx91//eE0ElC/89iY 7GkL0tDasmVx4TOXnrqrsziUcxEPPqHRE8x4NhtBK7+8o0nsMIJMA1gyZ2FA5To2Ew1BBuvovvDJ +Nua3qOCNBNu+8A+eNpJlVnlu/qB7+XWaPXtUMlsIikxD+gREFVUgYE4VzBuLa2kkg0VLd09XkE2 ceRDm6YgRATuDk6ogUyX4OLxCGIJF8yi6Z37M0wemDA6Uff0EuqdwDQd5HwG/rernUjt1grLdAxq 8BwywRRg0eFHmE+ShhpyO3Fi+wIDAQABo4HlMIHiMA4GA1UdDwEB/wQEAwIBBjASBgNVHRMBAf8E CDAGAQH/AgEAMB0GA1UdDgQWBBQ/FdJtfC/nMZ5DCgaolGwsO8XuZTBHBgNVHSAEQDA+MDwGBFUd IAAwNDAyBggrBgEFBQcCARYmaHR0cHM6Ly93d3cuZ2xvYmFsc2lnbi5jb20vcmVwb3NpdG9yeS8w MwYDVR0fBCwwKjAooCagJIYiaHR0cDovL2NybC5nbG9iYWxzaWduLm5ldC9yb290LmNybDAfBgNV HSMEGDAWgBRge2YaRQ2XyolQL30EzTSo//z9SzANBgkqhkiG9w0BAQUFAAOCAQEAQ3N5zKTMSTED HGFAgd/gu91Kb8AxPHgjq+7dhf7mkCinMqqrLai2XOrz8CP63BPaAx7oGOUBI0MyASBGk5zej9L3 oHtiF2BL01m1sBnT8rQxT2CJd/+jqjUl0p2ew8T3HSyatrsooGvDwf00yCB2JHTNvtQxNO8t6x/+ 048A1Q+0i7uf0nTnyrJLjD04zhL89ytetZspltOpJVYbmwiFjq6PxsdUNthUDme/9pOLmKDnQU0p W/JEwLs2TYCBNKwdgSGAk8/z+s2SCltKIG0Uh5U6t6j7JPuwNP/znImwMrlHDJ1YpW0rkF2PGraV CgDBf9dOB+IIpnwHfIi+LD+eITCCBMowggOyoAMCAQICEQCWaWbA3qWpL+Qmn6I16DynMA0GCSqG SIb3DQEBBQUAMFQxCzAJBgNVBAYTAkJFMRkwFwYDVQQKExBHbG9iYWxTaWduIG52LXNhMSowKAYD VQQDEyFHbG9iYWxTaWduIFBlcnNvbmFsU2lnbiAyIENBIC0gRzIwHhcNMTMwODI3MTY1NzU4WhcN MTYwODI3MTY1NzU4WjBYMQswCQYDVQQGEwJERTEcMBoGA1UEAxMTRGFuaWVsIEhlbGdlbmJlcmdl cjErMCkGCSqGSIb3DQEJARYcZGFuaWVsLmhlbGdlbmJlcmdlckBtLWJveC5kZTCCASIwDQYJKoZI hvcNAQEBBQADggEPADCCAQoCggEBAM4BQ5vPknk1OGLd1qKSUIKmQLrjccjJcYj7qtAtA+fNYKF8 9p1VY4UwiFcF9jKlmA9Q8o8tYSx16LYYFoGWokNRAeKFXZiBZiHyI0ekpEfxo8N5cTMCcxKcSYWV 8sqzmBPCoMNpmiVoC8ec8Nv5SqXH34VVtDmNLfiVlsTyomBXAJkJ2/n5XqJzPLFGWWREtPLkVVS+ u426vt/hNsQi5akNoidYeXo98JcrmeApFJ3zB2KxvMziHx8LD4q1gAl9NumtX5YLbCpdWL9AkWdX Oaro3D9zj6Q6LyGwa/UQUrZdg3BXc07hjHZn6d9vet1SzpbyqQpTzM63yXiX1meEMlMCAwEAAaOC AZEwggGNMA4GA1UdDwEB/wQEAwIFoDBMBgNVHSAERTBDMEEGCSsGAQQBoDIBKDA0MDIGCCsGAQUF BwIBFiZodHRwczovL3d3dy5nbG9iYWxzaWduLmNvbS9yZXBvc2l0b3J5LzAnBgNVHREEIDAegRxk YW5pZWwuaGVsZ2VuYmVyZ2VyQG0tYm94LmRlMAkGA1UdEwQCMAAwHQYDVR0lBBYwFAYIKwYBBQUH AwIGCCsGAQUFBwMEMEMGA1UdHwQ8MDowOKA2oDSGMmh0dHA6Ly9jcmwuZ2xvYmFsc2lnbi5jb20v Z3MvZ3NwZXJzb25hbHNpZ24yZzIuY3JsMFUGCCsGAQUFBwEBBEkwRzBFBggrBgEFBQcwAoY5aHR0 cDovL3NlY3VyZS5nbG9iYWxzaWduLmNvbS9jYWNlcnQvZ3NwZXJzb25hbHNpZ24yZzIuY3J0MB0G A1UdDgQWBBS8NFA/upd+Wipw2nj8RD/Ct+R2GTAfBgNVHSMEGDAWgBQ/FdJtfC/nMZ5DCgaolGws O8XuZTANBgkqhkiG9w0BAQUFAAOCAQEAXVTpu4fhOLETAW0zdbQiIwBIMZgeVNJnWV3GsMxByycU 63P+WBQTBl9qj47vHLmVdeF7MzH0QSXZSc9Tnfr6CYIImpyIZxRAGpAsWmtZf3JieRA0+j4GQJF2 zAea1NXYXoG9+ZSSZHBSxKUdrRdVdE320nuVGTT2HjEI2LEYbOvaXyi6HhpuHUiyu4LD0+RIT3fi T8jUiKKLTsApTD+Ak8SLF0IESOSA6htirv69mDDC7Klg9dT7QBPO7dpoKIUOldV3VhahndVfsDff KD7pkUUvG5XftYEQOxlWDJzuTBeqf/4hxXMtzFU9OaI6oKJjLfr6B+XBc6xwOtc/NMWmejCCBMow ggOyoAMCAQICEQCWaWbA3qWpL+Qmn6I16DynMA0GCSqGSIb3DQEBBQUAMFQxCzAJBgNVBAYTAkJF MRkwFwYDVQQKExBHbG9iYWxTaWduIG52LXNhMSowKAYDVQQDEyFHbG9iYWxTaWduIFBlcnNvbmFs U2lnbiAyIENBIC0gRzIwHhcNMTMwODI3MTY1NzU4WhcNMTYwODI3MTY1NzU4WjBYMQswCQYDVQQG EwJERTEcMBoGA1UEAxMTRGFuaWVsIEhlbGdlbmJlcmdlcjErMCkGCSqGSIb3DQEJARYcZGFuaWVs LmhlbGdlbmJlcmdlckBtLWJveC5kZTCCASIwDQYJKoZIhvcNAQEBBQADggEPADCCAQoCggEBAM4B Q5vPknk1OGLd1qKSUIKmQLrjccjJcYj7qtAtA+fNYKF89p1VY4UwiFcF9jKlmA9Q8o8tYSx16LYY FoGWokNRAeKFXZiBZiHyI0ekpEfxo8N5cTMCcxKcSYWV8sqzmBPCoMNpmiVoC8ec8Nv5SqXH34VV tDmNLfiVlsTyomBXAJkJ2/n5XqJzPLFGWWREtPLkVVS+u426vt/hNsQi5akNoidYeXo98JcrmeAp FJ3zB2KxvMziHx8LD4q1gAl9NumtX5YLbCpdWL9AkWdXOaro3D9zj6Q6LyGwa/UQUrZdg3BXc07h jHZn6d9vet1SzpbyqQpTzM63yXiX1meEMlMCAwEAAaOCAZEwggGNMA4GA1UdDwEB/wQEAwIFoDBM BgNVHSAERTBDMEEGCSsGAQQBoDIBKDA0MDIGCCsGAQUFBwIBFiZodHRwczovL3d3dy5nbG9iYWxz aWduLmNvbS9yZXBvc2l0b3J5LzAnBgNVHREEIDAegRxkYW5pZWwuaGVsZ2VuYmVyZ2VyQG0tYm94 LmRlMAkGA1UdEwQCMAAwHQYDVR0lBBYwFAYIKwYBBQUHAwIGCCsGAQUFBwMEMEMGA1UdHwQ8MDow OKA2oDSGMmh0dHA6Ly9jcmwuZ2xvYmFsc2lnbi5jb20vZ3MvZ3NwZXJzb25hbHNpZ24yZzIuY3Js MFUGCCsGAQUFBwEBBEkwRzBFBggrBgEFBQcwAoY5aHR0cDovL3NlY3VyZS5nbG9iYWxzaWduLmNv bS9jYWNlcnQvZ3NwZXJzb25hbHNpZ24yZzIuY3J0MB0GA1UdDgQWBBS8NFA/upd+Wipw2nj8RD/C t+R2GTAfBgNVHSMEGDAWgBQ/FdJtfC/nMZ5DCgaolGwsO8XuZTANBgkqhkiG9w0BAQUFAAOCAQEA XVTpu4fhOLETAW0zdbQiIwBIMZgeVNJnWV3GsMxByycU63P+WBQTBl9qj47vHLmVdeF7MzH0QSXZ Sc9Tnfr6CYIImpyIZxRAGpAsWmtZf3JieRA0+j4GQJF2zAea1NXYXoG9+ZSSZHBSxKUdrRdVdE32 0nuVGTT2HjEI2LEYbOvaXyi6HhpuHUiyu4LD0+RIT3fiT8jUiKKLTsApTD+Ak8SLF0IESOSA6hti rv69mDDC7Klg9dT7QBPO7dpoKIUOldV3VhahndVfsDffKD7pkUUvG5XftYEQOxlWDJzuTBeqf/4h xXMtzFU9OaI6oKJjLfr6B+XBc6xwOtc/NMWmejGCAucwggLjAgEBMGkwVDELMAkGA1UEBhMCQkUx GTAXBgNVBAoTEEdsb2JhbFNpZ24gbnYtc2ExKjAoBgNVBAMTIUdsb2JhbFNpZ24gUGVyc29uYWxT aWduIDIgQ0EgLSBHMgIRAJZpZsDepakv5CafojXoPKcwCQYFKw4DAhoFAKCCAVMwGAYJKoZIhvcN AQkDMQsGCSqGSIb3DQEHATAcBgkqhkiG9w0BCQUxDxcNMTQwNzE3MTIzMzI3WjAjBgkqhkiG9w0B CQQxFgQUgS34g4MSYT/5V1iCgFG2RuIBQPAweAYJKwYBBAGCNxAEMWswaTBUMQswCQYDVQQGEwJC RTEZMBcGA1UEChMQR2xvYmFsU2lnbiBudi1zYTEqMCgGA1UEAxMhR2xvYmFsU2lnbiBQZXJzb25h bFNpZ24gMiBDQSAtIEcyAhEAlmlmwN6lqS/kJp+iNeg8pzB6BgsqhkiG9w0BCRACCzFroGkwVDEL MAkGA1UEBhMCQkUxGTAXBgNVBAoTEEdsb2JhbFNpZ24gbnYtc2ExKjAoBgNVBAMTIUdsb2JhbFNp Z24gUGVyc29uYWxTaWduIDIgQ0EgLSBHMgIRAJZpZsDepakv5CafojXoPKcwDQYJKoZIhvcNAQEB BQAEggEAiM9uEmADg334zJt8QcmlqZ8FdWuBDT5btqwvpaQf47DRMjGRHQ2UkNZSvZ/nAahEXRSj CSzqnBeBFHQJ6D2E1p2Ixuns6/b4k7qyijLnoQ+EF3NisTr1foWfkHtw4sEarIKleUTfPR89O7gy BEkZXSxHXw+OSFvUiRd+ZquLdS5sNgPzNhZIv9kcTVYWcuiig4UhDab8Bh8jRpoimhOWJn7RTGrM ++b13/DlyGm3B2LM+8YVlkRTPkDUYQdSqc8/+Awky7e/dkWNW/4Bz2CcGMuwOnBS1ZVokZtvekTW eu59344FiYSIXvbtyYgotZfYqjsloSJa3vKB/UZylUIq8wAAAAAAAA== --=-aZsrVzyAkVB3GVWiu1xW--
Thanks for your thorough answer and explaination. I have gone with the direct lun to start with. I bought new Sata disks and added them to the storage chassi for the purpose of file sharing. Regards, Niklas
On 17 jul 2014, at 14:33, "Daniel Helgenberger" <daniel.helgenberger@m-box.de> wrote:
Hello,
for some reason the message was flagged spam; so there was a deplay.
oVirt supports direct LUNs. These LUNs are often already partitions of some RAID enclosure. AFAIK the MSA60 is a JBOD. You can use the p411 controller to create your partitions / LUNs.
The Virtio-SCSI paravirt driver supports a wide range of (=most) SCSI commands. This way clients can access them as 'real' SCSI devices.
If you did partition the LUN with parted, then the client(s) will see these partitions also, along with the file system on it.
As you might know, you cannot have 'normal' file systems mounted rw on several machines at one, you need a cluster file system; see examples in [1]; there are several open source FS's around.
Also, nothing stops you from mounting a file system read only on several hosts.
One note on the subject, though: I consider shared disk file systems as 'old' approach. I support them in our setup because of historical reasons. Today in a new deployment I would tend to use more 'modern' scale out file systems like GlusterFS (support in oVirt is quite well) or Ceph/Rados as object store file system. Using the native clients, you basically have a shared disk file system with less bottlenecks (MDC's in shared disk fs). Also, both examples have APIs - an application using this can greatly benefit in performance. Again, I consider using APIs for file storage the approach of the future.
If you need to attach NFS / CIFS clients, you can always reshare these file systems and use (p)NFS or CTDB if you want to cluster this.
Note, with one host and one JBOD this makes little sense to me.
[1] http://en.wikipedia.org/wiki/Clustered_file_system [2] https://ctdb.samba.org/
On Mo, 2014-07-14 at 08:10 +0000, Niklas Fondberg wrote: Thanks, after reading it makes sense. I suppose I need to drop my hope of having sharing possibility with the host. I have two questions that you might be able to answer: 1. Does direct LUN support partitions or only whole devices 2. Do you know of any open source way of making them shareable to Ovirt?
My setup is simple: - HP DL380 with dual Xeon and lots of RAM. Boots from seperate disk (USB) - MSA60 with p411 attached to the HP DL380
When we grow oVirt we will grow with DL360¹s and use all shared storage from the guest fileserver on the first host.
On 14/07/14 08:57, "Daniel Helgenberger" <daniel.helgenberger@m-box.de> wrote:
Hello,
just add my 2ct: I did a lot of bench marking for our SAN (FC LUN's). I also need file servers for our SMB Clients.
I recommend using Direct Attached LUNs for your purpose and attach them to the VMs as VirtIO-SCSI disks. You can even added them as shareable to oVirt if you deploy some kind of SAN file system (we use Quantum's StorNext).
Bottom line, the implementation of VirtIO-SCSI is so well done and support in oVirt is great. I cound not see any bottlenecks in the visualization. For the foreseeable future I will not deploy bare metal file Servers again.
HTH,
On So, 2014-07-13 at 15:47 +0000, Niklas Fondberg wrote:
From: Karli Sjöberg <Karli.Sjoberg@slu.se<mailto:Karli.Sjoberg@slu.se>> Date: Sunday 13 July 2014 14:51 To: Niklas Fondberg <niklas@vireone.com<mailto:niklas@vireone.com>> Cc: "users@ovirt.org<mailto:users@ovirt.org>" <users@ovirt.org<mailto:users@ovirt.org>>, Karli Sjöberg <Karli.Sjoberg@slu.se<mailto:Karli.Sjoberg@slu.se>> Subject: Re: [ovirt-users] fileserver as a guest oVirt
Den 12 jul 2014 22:49 skrev Niklas Fondberg <niklas@vireone.com<mailto:niklas@vireone.com>>:
On 12 jul 2014, at 16:57, "Karli Sjöberg" <Karli.Sjoberg@slu.se<mailto:Karli.Sjoberg@slu.se>> wrote:
Den 12 jul 2014 15:45 skrev Niklas Fondberg
> > Hi, > > I'm new to oVirt but I must say I am impressed! > I am running it on a HP DL380 with an external SAS chassi. > Linux dist is Centos 6.5 and oVirt is 3.4 running all-in-one (for now until we need to have a second host). > > Our company (www.vireone.com) deals with system architecture for many telco and media operators and is now setting up a small own datacenter for our internal tests as well as our IT infrastructure. > We are in the process of installing Zentyal for the SMB purposes on a guest and it would be great to have that guest also serving a fs
> > Does anyone have an idea of how I can through oVirt (seen several solutions using virsh and kvm) letting my Zentyal Ubuntu guest have access to a host mount point or if necessary (second best) a seperate
<niklas@vireone.com<mailto:niklas@vireone.com>>: path directory with NFS + SMB (which is semi crippled on the host after oVirt installation with version 3 et.c.). partition?
> > Best regards > Niklas
Why not just give the guest a thin provision virtual hard drive and expand it on a demand basis?
/K
Thanks for the advise but this would not suite us I'm afraid. It would be difficult wrt incremental backups as well as host machine file-routines.
Well, going by Occam's raizor; the simplest answer is usually correct. Can't really tell what you mean by file-routines but backups would be well served by snapshots (can't get more incremental than that) and disaster recovery could be as easy as a rsync from inside the guest to a remote machine.
The biggest pros here is the ease of being able to setup an export domain, attach, export the VM, detach domain, and then attach and import to a "real" setup when the AIO starts feeling crowded later on. Thinking ahead is never a bad thing, no?
/K
Thanks for your suggestions! The thing also is that the performance will be very bad if we have the 25TB SAS array shared for our purposes (lots of media streaming) using a virtual disk. What I am after (after more reading) is support for virtio-9p-pci (http://www.linux-kvm.org/page/9p_virtio) using oVirt. Alternative is the Direct LUN hook (http://www.ovirt.org/VDSM-Hooks/directlun, if I can figure out how to work with hooks...) Any chance anybody has an answer for these questions?
_______________________________________________ Users mailing list Users@ovirt.org http://lists.ovirt.org/mailman/listinfo/users
--
Daniel Helgenberger m box bewegtbild GmbH
P: +49/30/2408781-22 F: +49/30/2408781-10
ACKERSTR. 19 D-10115 BERLIN
www.m-box.de www.monkeymen.tv
Geschäftsführer: Martin Retschitzegger / Michaela Göllner Handeslregister: Amtsgericht Charlottenburg / HRB 112767
--
Daniel Helgenberger m box bewegtbild GmbH
P: +49/30/2408781-22 F: +49/30/2408781-10
ACKERSTR. 19 D-10115 BERLIN
www.m-box.de www.monkeymen.tv
Geschäftsführer: Martin Retschitzegger / Michaela Göllner Handeslregister: Amtsgericht Charlottenburg / HRB 112767
participants (3)
-
Daniel Helgenberger -
Karli Sjöberg -
Niklas Fondberg