FreeBSD 8.3
I just set up a system designed to handle lots of VBox VM's: a 6-core SandyBridge processor with 32GB of ram on a FreeBSD 8.3-STABLE host. Unfortunately, once I got it set up, I found that VBox guest performance simply sucks. The *mouse* isn't responsive. Linux guests see lots of CPU Locked errors. Googling turns up lots of problems with VBox on SandyBridge: Mac's running on 64 bit kernels have this problem - running on 32 bit kernels helps. Turning off Intel's SpeedStep has been reported to help. Turning off VT-X in the guest - if it's a 32-bit guest - helps. The last one is the only one that actually helped me. I was wondering if anyone here had run into and solved similar problems? Or if they were running a similar system, and didn't run into that problem? Especially anyone running 9.0 instead of 8.X! Thanks, mike -- Mike Meyer m...@mired.org http://www.mired.org/ Independent Software developer/SCM consultant, email for more information. O ascii ribbon campaign - stop html mail - www.asciiribbon.org ___ freebsd-hackers@freebsd.org mailing list http://lists.freebsd.org/mailman/listinfo/freebsd-hackers To unsubscribe, send any mail to freebsd-hackers-unsubscr...@freebsd.org
Re: FreeBSD 8.3
SandyBridge processor with 32GB of ram on a FreeBSD 8.3-STABLE host. i am using virtualbox but no linux guest, only windows. the newest CPU i have is CPU: Intel(R) Xeon(R) CPU E31220 @ 3.10GHz (3100.03-MHz K8-class CPU) Unfortunately, once I got it set up, I found that VBox guest performance simply sucks. The *mouse* isn't responsive. Linux guests see lots of CPU Locked errors. FreeBSD have linux emulation. i think you should use it instead of VBox if you need to run linux environments under FreeBSD. jails are your friend too. Windows runs great (as for windows of course) under VBox. ___ freebsd-hackers@freebsd.org mailing list http://lists.freebsd.org/mailman/listinfo/freebsd-hackers To unsubscribe, send any mail to freebsd-hackers-unsubscr...@freebsd.org
Re: FreeBSD 8.3
On Sat, 14 Jul 2012 19:56:22 +0200 (CEST) Wojciech Puchar woj...@wojtek.tensor.gdynia.pl wrote: CPU: Intel(R) Xeon(R) CPU E31220 @ 3.10GHz (3100.03-MHz K8-class CPU) If that the one normally listed as E3-1220? If so, it's a Sandy Bridge processor. If not, then I have not idea what it is. Unfortunately, once I got it set up, I found that VBox guest performance simply sucks. The *mouse* isn't responsive. Linux guests see lots of CPU Locked errors. FreeBSD have linux emulation. i think you should use it instead of VBox if you need to run linux environments under FreeBSD. Yes, it does. And I use it when I can. However, there are applications that it won't run, because of missing kernel features. And of course, it does absolutely not good at all if you need to run something other than Linux. Sucky Linux emulation performance is not my problem. If it were, I would have asked about that. My problem is sucky virtualbox performance, on any guest that has VT-X emulation enabled (which means all 64 bit guests). Windows runs great (as for windows of course) under VBox. Not for me. *Every* 64-bit guest OS has sucky performance. Linux, Windows, and others. The only one that reports any problems is Linux - it reports the CPU Locked errors. The others just suck. If I cut the memory size down and create 32-bit guests, they all seem to run OK. But I want to run systems for which there aren't 32-bit distributions. Are you only running 32-bit guests? Are you running on 9.x or 8.x? Thanx, mike -- Mike Meyer m...@mired.org http://www.mired.org/ Independent Software developer/SCM consultant, email for more information. O ascii ribbon campaign - stop html mail - www.asciiribbon.org ___ freebsd-hackers@freebsd.org mailing list http://lists.freebsd.org/mailman/listinfo/freebsd-hackers To unsubscribe, send any mail to freebsd-hackers-unsubscr...@freebsd.org
Re: FreeBSD 8.3
On Sat, Jul 14, 2012 at 06:11:41AM -0400, Mike Meyer wrote: I just set up a system designed to handle lots of VBox VM's: a 6-core SandyBridge processor with 32GB of ram on a FreeBSD 8.3-STABLE host. Unfortunately, once I got it set up, I found that VBox guest performance simply sucks. The *mouse* isn't responsive. Linux guests see lots of CPU Locked errors. Try booting the linux VBox with HZ=100. This should greatly improve your performance. Googling turns up lots of problems with VBox on SandyBridge: Mac's running on 64 bit kernels have this problem - running on 32 bit kernels helps. Turning off Intel's SpeedStep has been reported to help. Turning off VT-X in the guest - if it's a 32-bit guest - helps. The last one is the only one that actually helped me. I was wondering if anyone here had run into and solved similar problems? Or if they were running a similar system, and didn't run into that problem? Especially anyone running 9.0 instead of 8.X! Thanks, mike -- Mike Meyer m...@mired.org http://www.mired.org/ Independent Software developer/SCM consultant, email for more information. O ascii ribbon campaign - stop html mail - www.asciiribbon.org ___ freebsd-hackers@freebsd.org mailing list http://lists.freebsd.org/mailman/listinfo/freebsd-hackers To unsubscribe, send any mail to freebsd-hackers-unsubscr...@freebsd.org -- - (2^(N-1)) ___ freebsd-hackers@freebsd.org mailing list http://lists.freebsd.org/mailman/listinfo/freebsd-hackers To unsubscribe, send any mail to freebsd-hackers-unsubscr...@freebsd.org
Re: FreeBSD 8.3
Wojciech Puchar woj...@wojtek.tensor.gdynia.pl wrote: CPU: Intel(R) Xeon(R) CPU E31220 @ 3.10GHz (3100.03-MHz K8-class CPU) If that the one normally listed as E3-1220? If so, it's a Sandy Bridge yes. processor. If not, then I have not idea what it is. so it is. it have working AES-NI. I am such a kind of person that i am completely not on-time with marketing, namings, etc.. I just asked what CPUs will support AES-NI which is important for me (geli speedup) and got that in Dell Server ;) Virtualbox works great with windoze on that machine. FreeBSD have linux emulation. i think you should use it instead of VBox if you need to run linux environments under FreeBSD. Yes, it does. And I use it when I can. However, there are applications that it won't run, because of missing kernel features. And of course, it does absolutely not good at all if you need to run something other than Linux. would have asked about that. My problem is sucky virtualbox performance, on any guest that has VT-X emulation enabled (which means all 64 bit guests). i would rather bet on linux addon kernel modules you have to install. In windows you have to install guest additions without this it is plain terrible. Windows runs great (as for windows of course) under VBox. Not for me. *Every* 64-bit guest OS has sucky performance. Linux, Are you sure with two level pagetables featured in modern CPUs including sandy bridge? Are you only running 32-bit guests? Are you running on 9.x or 8.x? for production - yes (6 instances of windows XP). Virtualbox does never make main workload for me, it is just addon. On FreeBSD 8.3. But i've tried Windows 7 64-bit and it worked fine. didn't see any problems with latency you describe ___ freebsd-hackers@freebsd.org mailing list http://lists.freebsd.org/mailman/listinfo/freebsd-hackers To unsubscribe, send any mail to freebsd-hackers-unsubscr...@freebsd.org
Re: FreeBSD 8.3
On Sat, 14 Jul 2012 20:27:11 +0200 (CEST) Wojciech Puchar woj...@wojtek.tensor.gdynia.pl wrote: would have asked about that. My problem is sucky virtualbox performance, on any guest that has VT-X emulation enabled (which means all 64 bit guests). i would rather bet on linux addon kernel modules you have to install. Are you talking about the guest additions? They're already installed (on a VM that was running on an older Core 2 CPU). Performance sucks. In windows you have to install guest additions without this it is plain terrible. I haven't managed to get through an install on a 64-bit windows system yet to try that. However, that Linux doesn't Linux, or for the 32-bit Windows runs great (as for windows of course) under VBox. Not for me. *Every* 64-bit guest OS has sucky performance. Linux, Are you sure with two level pagetables featured in modern CPUs including sandy bridge? Yes, I'm sure that every guest OS I've tried on a 64 bit guest sucks. I'm busy recreating 32-bit versions of the 64-bit guests where I can. Are you only running 32-bit guests? Are you running on 9.x or 8.x? for production - yes (6 instances of windows XP). Virtualbox does never make main workload for me, it is just addon. On FreeBSD 8.3. Could you tell me if they all have VT-X disabled? But i've tried Windows 7 64-bit and it worked fine. didn't see any problems with latency you describe Could you send me the system settings (VT-X, PAE, etc.) you used for this? Thanks, mike -- Mike Meyer m...@mired.org http://www.mired.org/ Independent Software developer/SCM consultant, email for more information. O ascii ribbon campaign - stop html mail - www.asciiribbon.org ___ freebsd-hackers@freebsd.org mailing list http://lists.freebsd.org/mailman/listinfo/freebsd-hackers To unsubscribe, send any mail to freebsd-hackers-unsubscr...@freebsd.org
Re: FreeBSD 8.3
On Sat, 14 Jul 2012, Mike Meyer wrote: I just set up a system designed to handle lots of VBox VM's: a 6-core SandyBridge processor with 32GB of ram on a FreeBSD 8.3-STABLE host. Unfortunately, once I got it set up, I found that VBox guest performance simply sucks. The *mouse* isn't responsive. Linux guests see lots of CPU Locked errors. Can you give a specific Linux version that has problems? I'm willing to download and test it on this i5/9-stable/amd64 system. Haven't noticed any problems, but I only occasionally boot Ubuntu in a VM for a little bit. ___ freebsd-hackers@freebsd.org mailing list http://lists.freebsd.org/mailman/listinfo/freebsd-hackers To unsubscribe, send any mail to freebsd-hackers-unsubscr...@freebsd.org
Re: FreeBSD 8.3
On Sat, 14 Jul 2012 13:13:50 -0600 (MDT) Warren Block wbl...@wonkity.com wrote: On Sat, 14 Jul 2012, Mike Meyer wrote: Can you give a specific Linux version that has problems? I'm willing to download and test it on this i5/9-stable/amd64 system. Haven't noticed any problems, but I only occasionally boot Ubuntu in a VM for a little bit. 64-bit Ubuntu LTS 12.04. I moved a VM from the previous system, where it worked fine (same build of FreeBSD, same build of VirtualBox). The OS seems to be irrelevant. Windows XP and 7 and mumble all have this problem, *if* I have VT-X enabled in VirtualBox. If I disable VT-X, the ones I have tested so far worked fine. I'm still getting 32-bit builds of some of them, as you can't turn VT-X off in a 64-bit guest. Thanks, mike -- Mike Meyer m...@mired.org http://www.mired.org/ Independent Software developer/SCM consultant, email for more information. O ascii ribbon campaign - stop html mail - www.asciiribbon.org ___ freebsd-hackers@freebsd.org mailing list http://lists.freebsd.org/mailman/listinfo/freebsd-hackers To unsubscribe, send any mail to freebsd-hackers-unsubscr...@freebsd.org
Re: FreeBSD 8.3
On Sat, 14 Jul 2012, Mike Meyer wrote: On Sat, 14 Jul 2012 13:13:50 -0600 (MDT) Warren Block wbl...@wonkity.com wrote: On Sat, 14 Jul 2012, Mike Meyer wrote: Can you give a specific Linux version that has problems? I'm willing to download and test it on this i5/9-stable/amd64 system. Haven't noticed any problems, but I only occasionally boot Ubuntu in a VM for a little bit. 64-bit Ubuntu LTS 12.04. I moved a VM from the previous system, where it worked fine (same build of FreeBSD, same build of VirtualBox). The OS seems to be irrelevant. Windows XP and 7 and mumble all have this problem, *if* I have VT-X enabled in VirtualBox. If I disable VT-X, the ones I have tested so far worked fine. I'm still getting 32-bit builds of some of them, as you can't turn VT-X off in a 64-bit guest. In a VM with stock settings (Linux 64-bit), Ubuntu 12.04 LTS Desktop Live CD works okay, installed also seems to work okay. The mouse is a little draggy but usable, like using a wireless mouse. kern.hz is set to 100 on the host (9-stable, er 9.1-BETA1, amd64). This is without guest additions. Trying the VirtualBox menu install guest additions made it crash. The only mouse problem I saw was the mouse pointer disappearing when over some preferences icons. ___ freebsd-hackers@freebsd.org mailing list http://lists.freebsd.org/mailman/listinfo/freebsd-hackers To unsubscribe, send any mail to freebsd-hackers-unsubscr...@freebsd.org
Re: FreeBSD 8.3
On Sat, Jul 14, 2012 at 2:27 PM, Mike Meyer m...@mired.org wrote: 64-bit Ubuntu LTS 12.04. I moved a VM from the previous system, where it worked fine (same build of FreeBSD, same build of VirtualBox). The OS seems to be irrelevant. Windows XP and 7 and mumble all have this problem, *if* I have VT-X enabled in VirtualBox. If I disable VT-X, the ones I have tested so far worked fine. I'm still getting 32-bit builds of some of them, as you can't turn VT-X off in a 64-bit guest. If possible, set VM to single cpu. Also not sure how you migrated machines. Occasionally the VM export/import functionality has produced silliness. Try creating new VM from scratch then attaching existing VM disk(s) to it. -- Adam Vande More ___ freebsd-hackers@freebsd.org mailing list http://lists.freebsd.org/mailman/listinfo/freebsd-hackers To unsubscribe, send any mail to freebsd-hackers-unsubscr...@freebsd.org
Re: FreeBSD 8.3
For the OP, make sure you have the latest BIOS. I had a similar problem with vt-x and it was solved by a later BIOS upgrade. hth, Doug -- Change is hard. ___ freebsd-hackers@freebsd.org mailing list http://lists.freebsd.org/mailman/listinfo/freebsd-hackers To unsubscribe, send any mail to freebsd-hackers-unsubscr...@freebsd.org
Re: FreeBSD 8.3
On Sat, 14 Jul 2012, Warren Block wrote: In a VM with stock settings (Linux 64-bit), Ubuntu 12.04 LTS Desktop Live CD works okay, installed also seems to work okay. The mouse is a little draggy but usable, like using a wireless mouse. kern.hz is set to 100 on the host (9-stable, er 9.1-BETA1, amd64). This is without guest additions. Trying the VirtualBox menu install guest additions made it crash. The only mouse problem I saw was the mouse pointer disappearing when over some preferences icons. Switched to XUbuntu because the menus were irritating in plain Ubuntu. No serious problems there either, and at least the package manager is locatable. ___ freebsd-hackers@freebsd.org mailing list http://lists.freebsd.org/mailman/listinfo/freebsd-hackers To unsubscribe, send any mail to freebsd-hackers-unsubscr...@freebsd.org
Re: FreeBSD 8.3
i would rather bet on linux addon kernel modules you have to install. Are you talking about the guest additions? They're already installed (on a VM that was running on an older Core 2 CPU). Performance sucks. something must be wrong with linux interacting with virtualbox. Won't help you as i never used linux on it, and actually at all for long time. In windows you have to install guest additions without this it is plain terrible. I haven't managed to get through an install on a 64-bit windows system Windows 7 64-bit installs fine. FreeBSD 8.3, vbox 4.0.12 sandy bridge? Yes, I'm sure that every guest OS I've tried on a 64 bit guest sucks. I'm busy recreating 32-bit versions of the 64-bit guests where I can. maybe. i didn't do very detailed tests. and don't use 64-bit guest in production. problems with latency you describe Could you send me the system settings (VT-X, PAE, etc.) you used for this? Everything that is possible enable in BIOS, FreeBSD/amd64 8.3-stable (like month old or so), virtualbox 4.0.12 no fancy tricks, custom kernel but nothing special. If you need more ask on priv. ___ freebsd-hackers@freebsd.org mailing list http://lists.freebsd.org/mailman/listinfo/freebsd-hackers To unsubscribe, send any mail to freebsd-hackers-unsubscr...@freebsd.org
Re: FreeBSD 8.3
On Sa., 14. Jul. 2012 12:11:41 CEST, Mike Meyer m...@mired.org wrote: I just set up a system designed to handle lots of VBox VM's: a 6-core SandyBridge processor with 32GB of ram on a FreeBSD 8.3-STABLE host. Unfortunately, once I got it set up, I found that VBox guest performance simply sucks. The *mouse* isn't responsive. Linux guests see lots of CPU Locked errors. Does the VM use multiple vCPUs? Does reducing to one vcpu help? I have seen a lot of instabilities and problems in the past when using more than one vcpu. Where is that cpu locked reported? You could also try to build the vbox port with debug option enabled and look at VBox.log which gives probably useful information in case there are soft errors that cause the slowdown or if some of the vt-x features were incorrectly detected and vbox had to fallback to the recompiler (=slow). Googling turns up lots of problems with VBox on SandyBridge: Mac's running on 64 bit kernels have this problem - running on 32 bit kernels helps. Turning off Intel's SpeedStep has been reported to help. Turning off VT-X in the guest - if it's a 32-bit guest - helps. The last one is the only one that actually helped me. Just to be sure which vbox version are we talking about? You could try emulators/virtualbox-ose-legacy to verify if it is a regression of 4.1.x. I was wondering if anyone here had run into and solved similar problems? Or if they were running a similar system, and didn't run into that problem? Especially anyone running 9.0 instead of 8.X I've not heard about that before so it wasn't reported at least. ___ freebsd-hackers@freebsd.org mailing list http://lists.freebsd.org/mailman/listinfo/freebsd-hackers To unsubscribe, send any mail to freebsd-hackers-unsubscr...@freebsd.org
Interesting facts about AI_ADDRCONFIG
Hi, hackers: FreeBSD's AI_ADDRCONFIG flag used by getaddrinfo(3) is regarded as broken: https://wikispaces.psu.edu/display/ipv6/IPv6+programming#IPv6programming-AIADDRCONFIGflag The short reason is that it still queries for both A and even one of them is not configured. The details are interesting: In rfc2553, AI_ADDRCONFIG is described as: - The AI_ADDRCONFIG flag specifies that a query for records should occur only if the node has at least one IPv6 source address configured and a query for A records should occur only if the node has at least one IPv4 source address configured. Note that the description applies to getipnodebyname() actually, but we can expect that it also applies to getaddrinfo(). And compare the above with the description from POSIX.1-2008: If the AI_ADDRCONFIG flag is specified, IPv4 addresses shall be returned only if an IPv4 address is configured on the local system, and IPv6 addresses shall be returned only if an IPv6 address is configured on the local system. See the difference? POSIX does not care whether the addresses are queried; it only says that it should be not returned. Hence, you can query them first and filter them latter... However, this is not the biggest problem. The main chaos is about whether a loopback IP address can be considered configured: An embarrassing workaround by Mozilla: https://mxr.mozilla.org/mozilla-central/source/nsprpub/pr/src/misc/prnetdb.c#2013 In 2003, rfc3493 responded the problem as: - If the AI_ADDRCONFIG flag is specified, IPv4 addresses shall be returned only if an IPv4 address is configured on the local system, and IPv6 addresses shall be returned only if an IPv6 address is configured on the local system. The loopback address is not considered for this case as valid as a configured address. For example, when using the DNS, a query for records should occur only if the node has at least one IPv6 address configured (other than IPv6 loopback) and a query for A records should occur only if the node has at least one IPv4 address configured (other than the IPv4 loopback). Now I think things becomes clear: To support AI_ADDRCONFIG in a POSIX+RFC way, getaddrinfo(3) should detect outbound address availability first, then query. Let's go back to the code in FreeBSD: In lib/libc/net/getaddrinfo.c and lib/libc/net/name6.c , AI_ADDRCONFIG is implemented as filter out the unsupported AFs in kernel, then query. Obviously it's a misinterpretation of the word configured, and the implementation does not confirm with any standards. A better impl may be bind9's lwres_getipnodebyname() (in contrib/bind9/lib/lwres/getipnode.c). It carefully uses ioctl(2) to obtain the addr info on interfaces, then do query. I'm not sure whether it handles the loopback problem, but it's much better then what we have in libc. So... What we going to do? -- Zhihao Yuan, nickname lichray The best way to predict the future is to invent it. ___ 4BSD -- http://4bsd.biz/ ___ freebsd-hackers@freebsd.org mailing list http://lists.freebsd.org/mailman/listinfo/freebsd-hackers To unsubscribe, send any mail to freebsd-hackers-unsubscr...@freebsd.org
