suggested vhost link speed settings
Hi list, Being that the virtio interfaces are stated as acheiving 5-8 Gb throughput now with vhost, as opposed to 1Gb without, how should their link speed be defined when the choices are 2500M or 1M? I have them plotted out to make a 10Gb bond out of a pair, counting on 5Gb max each, which I'm imagining can be acheived without concern based on what I read. If I set them to 'auto-negotiate' will it internally flap between cause undesirable consequences? I don't want to set them at 2500 in case they ever do need to reach closer to 5 each. I'm not at a point where I can test anything yet, I'm just planning preconfiguring so far. For the sake of preventing issues down the line I wanted to see if there's a consensus or standard for this scenario and be as sure as possible ahead of time. Thanks, -C -- To unsubscribe from this list: send the line unsubscribe kvm in the body of a message to majord...@vger.kernel.org More majordomo info at http://vger.kernel.org/majordomo-info.html
Re: [RFC] GPGPU Support In KVM
Interesting, indeed. Looking forward to it as well. On Wed, 24 Nov 2010 20:56 +0100, André Weidemann andre.weidem...@web.de wrote: Hi, On 24.11.2010 15:06, Prasad Joshi wrote: I have been following the KVM mailing list last few months and have learned that the KVM does not have the GPU pass-through support. As far as I can understand, adding GPU pass-through would make GPU device available to VM as a Graphics Card, let me know if I am wrong. After the completion GPGPU support in VM I would love to work on this support as well. Please let me know your thoughts. There is already someone who was working on GPU pass-trough for KVM. Search the mailing list archives for _Fede_. Back in June he said that he would need to debug a BIOS issue in order to make it work. Maybe you two should get together. I hope this helps. I am really looking forward to see this in KVM. Regards André -- To unsubscribe from this list: send the line unsubscribe kvm in the body of a message to majord...@vger.kernel.org More majordomo info at http://vger.kernel.org/majordomo-info.html -- To unsubscribe from this list: send the line unsubscribe kvm in the body of a message to majord...@vger.kernel.org More majordomo info at http://vger.kernel.org/majordomo-info.html
Re: limit conectivity of a VM
if you had infinitely fast processors, every virtual network would be infinitely fast. I see on a Vyatta VM, that an interface's link speed attribute can be explicitly defined, along with duplex. Possible values are 10 100 1000 Mb, and are configured independently of the driver/model of NIC. I haven't tested it yet, and since discovering this detail, have been somewhat disheartened at the thought of ~8 Gb vhost throughput being throttled by the highest possible link speed setting being 1000 Mb. So maybe plan b could be to install a test router just for that function, and loop through it. -C On Sat, 20 Nov 2010 10:39 -0500, Javier Guerra Giraldez jav...@guerrag.com wrote: On Sat, Nov 20, 2010 at 3:40 AM, Thomas Mueller tho...@chaschperli.ch wrote: maybe one of the virtual network cards is 10mbit? start kvm with -net nic,model=? to get a list. wouldn't matter. different models emulate the hardware registers used to transmit, not the performance. if you had infinitely fast processors, every virtual network would be infinitely fast. -- Javier -- To unsubscribe from this list: send the line unsubscribe kvm in the body of a message to majord...@vger.kernel.org More majordomo info at http://vger.kernel.org/majordomo-info.html -- To unsubscribe from this list: send the line unsubscribe kvm in the body of a message to majord...@vger.kernel.org More majordomo info at http://vger.kernel.org/majordomo-info.html
New @ Proxmox: -device, vhost... Docs, notes?
Hi Everyone, I'm impressed with all the activity I see here since joining the list this year. It helps to reinforce that I chose the right technology. Thanks. The -device method vhost=on option recently became available to us at the ProxmoxVE project I'm preparing to start making use of them this coming week. I found some docs on linux-kvm.org, and still have a bit to look through see what I can find. I don't want to state on the record that I will or I won't but it's crossed my mind to assemble something for our wiki, and if I do decide to it would be nice to have some fresh material. So I'm just writing to see if anyone would like to share any of your favorite reference material- ie. links, benchmarks, drawings, diagrams, etc. Kind Regards, -C -- To unsubscribe from this list: send the line unsubscribe kvm in the body of a message to majord...@vger.kernel.org More majordomo info at http://vger.kernel.org/majordomo-info.html
Re: [v3 RFC PATCH 0/4] Implement multiqueue virtio-net
On Fri, 29 Oct 2010 13:26 +0200, Michael S. Tsirkin m...@redhat.com wrote: On Thu, Oct 28, 2010 at 12:48:57PM +0530, Krishna Kumar2 wrote: Krishna Kumar2/India/IBM wrote on 10/28/2010 10:44:14 AM: In practice users are very unlikely to pin threads to CPUs. I may be misunderstanding what you're referring to. It caught my attention since I'm working on a configuration to do what you say is unlikely, so I'll chime in for what it's worth. An option in Vyatta allows assigning CPU affinity to network adapters, since apparently seperate L2 caches can have a significant impact on throughput. Although much of their focus seems to be on commercial virtualization platforms, I do see quite a few forum posts with regard to KVM. Mabye this still qualifies as an edge case, but as for virtualized routing theirs seems to offer the most functionality. http://www.vyatta.org/forum/viewtopic.php?t=2697 -cb -- To unsubscribe from this list: send the line unsubscribe kvm in the body of a message to majord...@vger.kernel.org More majordomo info at http://vger.kernel.org/majordomo-info.html
Re: [Qemu-devel] Hitting 29 NIC limit (+Intel VT-c)
On Thu, 14 Oct 2010 14:07 +0200, Avi Kivity a...@redhat.com wrote: On 10/14/2010 12:54 AM, Anthony Liguori wrote: On 10/13/2010 05:32 PM, Anjali Kulkarni wrote: What's the motivation for such a huge number of interfaces? Ultimately to bring multiple 10Gb bonds into a Vyatta guest. --- BTW, I don't think it's possible to hot-add physical functions. I believe I know of a card that supports dynamic add of physical functions (pre-dating SR-IOV) I don't know what you're talking about, but it seems you have a better handle than I on this VT-c stuff, so perhaps misguidedly I'll direct my next question to you. Is additional configuration required to make use of SR-IOV VTq? I don't immediateley understand how the queueing knows who is who in the absense of eth.vlan- or if I need to for that matter. My hope is that this is something like plug n play as long as kernel, host driver versions are foo, but I haven't yet found documentation to confirm it. For the sake of future queries, I've come across these references so far: http://download.intel.com/design/network/applnots/321211.pdf http://www.linux-kvm.org/wiki/images/6/6a/KvmForum2008%24kdf2008_7.pdf http://www.mail-archive.com/kvm@vger.kernel.org/msg27860.html http://www.mail-archive.com/kvm@vger.kernel.org/msg22721.html http://thread.gmane.org/gmane.linux.kernel.mm/38508 http://ark.intel.com/Product.aspx?id=36918 -- To unsubscribe from this list: send the line unsubscribe kvm in the body of a message to majord...@vger.kernel.org More majordomo info at http://vger.kernel.org/majordomo-info.html
Re: NIC limit
The PCI bus has only 32 slots (devices), 3 taken by chipset + vga, and a 4th if you have, for example, a virtio disk. Are you sure these are 33 PCI devices and not 33 PCI functions? No, not sure. Apparently my statement was based on an uninformed assumption. I tested using a VM that had 30 (removable-per-web-interface) attachments, and added 3x IDE HDDs to bring it above what I thought was 32 devices: 28 virtio NICs 1 IDE CD-ROM 1 virtio HDD +3 IDE HDDs I could add IDE up past 32 and it would start, as soon as there were more than 28 NICs with or without the 3 IDE HDDs, start would fail. On Wed, 06 Oct 2010 10:18 -0700, Chris Wright chr...@sous-sol.org wrote: * linux_...@proinbox.com (linux_...@proinbox.com) wrote: Hi again everybody, One of the admins at the ProxmoxVE project was gracious enough to quickly release a package including the previously discussed change to allow up to 32 NICs in qemu. You mean they patched qemu to increase the MAX_NICS constant? Nice to get the quick turn around. Te better choice is to use a newer command line. Not only does it avoid the MAX_NICS limitation, but it also enables standard virtio-net offload accelerations. For future reference the .deb is here: ftp://download.proxmox.com/debian/dists/lenny/pvetest/binary-amd64/pve-qemu-kvm_0.12.5-2_amd64.deb Upon creating running the VM with the newly patched qemu-kvm app installed, I found a NIC limitation remained in place, presumably imposed by some other aspect of the environment. The machine would start when it had 33 PCI devices, as long as no more than 28 of them were NICs. The PCI bus has only 32 slots (devices), 3 taken by chipset + vga, and a 4th if you have, for example, a virtio disk. Are you sure these are 33 PCI devices and not 33 PCI functions? thanks, -chris -- To unsubscribe from this list: send the line unsubscribe kvm in the body of a message to majord...@vger.kernel.org More majordomo info at http://vger.kernel.org/majordomo-info.html
NIC limit
Hi again everybody, One of the admins at the ProxmoxVE project was gracious enough to quickly release a package including the previously discussed change to allow up to 32 NICs in qemu. For future reference the .deb is here: ftp://download.proxmox.com/debian/dists/lenny/pvetest/binary-amd64/pve-qemu-kvm_0.12.5-2_amd64.deb Upon creating running the VM with the newly patched qemu-kvm app installed, I found a NIC limitation remained in place, presumably imposed by some other aspect of the environment. The machine would start when it had 33 PCI devices, as long as no more than 28 of them were NICs. This is still a vast improvement compared to the previous limit of 8 NICs, and is very good news for my project. I post here in hopes that maybe someone will come across the link in a search and have a solution. More likely however the new API will be in place and widely in use by then, but whatever. Either way, thanks for your help yesterday. -- To unsubscribe from this list: send the line unsubscribe kvm in the body of a message to majord...@vger.kernel.org More majordomo info at http://vger.kernel.org/majordomo-info.html
Re: 8 NIC limit - patch - places limit at 32
It's 8 otherwise- and after the patch is applied, it still only goes to 28 for some reason. 28's acceptable for my needs, so I'll step aside from here leave it to the experts. As for the new -device method, that's all fine good but AFAIK it's not implemented on my platform, so this was the answer. On Wed, 06 Oct 2010 07:54 -0500, Anthony Liguori anth...@codemonkey.ws wrote: On 10/06/2010 12:46 AM, linux_...@proinbox.com wrote: Attached is a patch that allows qemu to have up to 32 NICs, without using the qdev -device method. I'd rather there be no fixed limit and we validate that when add fails because there isn't a TCP slot available, we do the right thing. BTW, using -device, it should be possible to add a very high number of nics because you can specify the PCI address including a function. If this doesn't Just Work today, we should make it work. Regards, Anthony Liguori -- To unsubscribe from this list: send the line unsubscribe kvm in the body of a message to majord...@vger.kernel.org More majordomo info at http://vger.kernel.org/majordomo-info.html
8 NIC limit
Hello list: I'm working on a project that calls for the creation of a firewall in KVM. While adding a 20-interface trunk of virtio adapters to bring in a dual 10GB bond, I've discovered an 8 NIC limit in QEMU. I found the following thread in the list archives detailing a similar problem: http://kerneltrap.org/mailarchive/linux-kvm/2009/1/29/4848304 It includes a patch for the file qemu/net.h to allow 24 NICs: https://bugs.launchpad.net/ubuntu/+source/qemu-kvm;qemu-kvm/+bug/595873/+attachment/1429544/+files/max_nics.patch In my case I want to attach 29, and have simply changed line 8 to 30 from 24. This will be the first patch I've ever had to do, and so far my internet search yields results that don't seem to apply. Would someone like to recommend a pertinent tutorial? Many thanks -- To unsubscribe from this list: send the line unsubscribe kvm in the body of a message to majord...@vger.kernel.org More majordomo info at http://vger.kernel.org/majordomo-info.html
Fwd: Re: 8 NIC limit
Forgot to cc list, forwarding. In this case, I think you're going to want to send your patch to the qemu-devel (on CC) mailing list (perhaps in addition to sending it here, to the kvm list). Will do, thanks for the pointer. Before I do so, I'd like to bring up one thing that comes to mind. I don't know how to make the determination, but it makes sense to me for the limit defined here to be indicitive of an actual limitation, rather than what seems an arbitrary best-guess as to the most someone might need. If the change ends up being permanent, then I would hope it would be a large enough value to provide a degree of extensibility and prevent the necessity of bumping it up again later when someone else comes along with even greater bandwidth requirements. Perhaps someone could provide some guidance as to a sane, higher number, as opposed to an arbitrary '65000' which would surely prevent this from happening again (knock on wood). For the time being I still have to find something to help learn how to implement the change locally. I rarely have to compile let alone deal with patches, so to me at least this is a considerable obstacle. -Thanks On Tue, 05 Oct 2010 08:24 -0700, Dustin Kirkland kirkl...@canonical.com wrote: On Tue, Oct 5, 2010 at 7:48 AM, linux_...@proinbox.com wrote: Hello list: I'm working on a project that calls for the creation of a firewall in KVM. While adding a 20-interface trunk of virtio adapters to bring in a dual 10GB bond, I've discovered an 8 NIC limit in QEMU. I found the following thread in the list archives detailing a similar problem: http://kerneltrap.org/mailarchive/linux-kvm/2009/1/29/4848304 It includes a patch for the file qemu/net.h to allow 24 NICs: https://bugs.launchpad.net/ubuntu/+source/qemu-kvm;qemu-kvm/+bug/595873/+attachment/1429544/+files/max_nics.patch In my case I want to attach 29, and have simply changed line 8 to 30 from 24. This will be the first patch I've ever had to do, and so far my internet search yields results that don't seem to apply. Would someone like to recommend a pertinent tutorial? Hi there, I commented on the original bug in Launchpad. We're willing and able to carry the patch against qemu-kvm in Ubuntu, I just asked that the reporter at least submit the patch upstream for discussion. I don't see where that has happened yet. It's a trivial patch to submit. Please note in that bug a pointer to the mailing list thread, if you start one. To your specific question, different communities have different requirements on patch submission, so you do need to consult each community. A good place to start might be the Documentation/SubmittingPatches how-to in the kernel tree: * http://git.kernel.org/?p=linux/kernel/git/torvalds/linux-2.6.git;a=blob_plain;f=Documentation/SubmittingPatches;hb=HEAD In this case, I think you're going to want to send your patch to the qemu-devel (on CC) mailing list (perhaps in addition to sending it here, to the kvm list). :-Dustin -- To unsubscribe from this list: send the line unsubscribe kvm in the body of a message to majord...@vger.kernel.org More majordomo info at http://vger.kernel.org/majordomo-info.html
Re: 8 NIC limit
Have you tried creating NICs with -device? I'm not sure what that is, will look into it, thanks. I'm using ProxmoxVE, and currently add them via a web interface. Someone happens to host a screenshot of that part here: http://c-nergy.be/blog/wp-content/uploads/Proxmox_Net2.png On Tue, 05 Oct 2010 17:57 +0200, Markus Armbruster arm...@redhat.com wrote: linux_...@proinbox.com writes: Hello list: I'm working on a project that calls for the creation of a firewall in KVM. While adding a 20-interface trunk of virtio adapters to bring in a dual 10GB bond, I've discovered an 8 NIC limit in QEMU. Have you tried creating NICs with -device? The limit shouldn't apply there. -- To unsubscribe from this list: send the line unsubscribe kvm in the body of a message to majord...@vger.kernel.org More majordomo info at http://vger.kernel.org/majordomo-info.html
Re: 8 NIC limit - patch - places limit at 32
Attached is a patch that allows qemu to have up to 32 NICs, without using the qdev -device method. max_nics.patch Description: Binary data