Re: IP forwarding performance

On Fri, Dec 14, 2012 at 5:47 PM, Sepherosa Ziehau sepher...@gmail.com wrote: Hi all, This email serves as the base performance measurement for further network stack optimization (as of git 107282b). Since bidirectional fast IP forwarding is already max out the GigE limit, I increase

IP forwarding performance (git.2fb36fa, normal 4.40Mpps, fast 5.23Mpps)

) +160Kpps for 2 bidirectional fast IP forwarding (now 5.23Mpps) During the performance measurement, the system is very responsive. Detailed information, please read the following inline comment. On Thu, Dec 27, 2012 at 4:42 PM, Sepherosa Ziehau sepher...@gmail.com wrote: Hi all, Before I move

IP forwarding performance, more improvement (git.8eb1b0, normal 4.61Mpps, fast 5.67Mpps)

fast IP forwarding (now 5.67Mpps) For fast IP forwarding, we are _not_ that far away from max out the 4 GigE interfaces (which is 5.95Mpps) Detailed information is same as last measurement. Best Regards, sephe On Fri, Feb 1, 2013 at 5:53 PM, Sepherosa Ziehau sepher...@gmail.com wrote: Hi all

Hofmann's wlan_serialize branch has been pushed

Hi all, I have pushed Hofmann's wlan_serialize branch, which solves many wlan related panics. however, it needs further testing on wpi(4) and iwi(4). If you have the hardware, please let's know whether it works for you or not. Best Regards, sephe -- Tomorrow Will Never Die

[HEADS UP] TSC cputimer is added

Hi all, For master users, TSC cputimer is added and enabled on x86_64 if MP synchronized TSC is detected. If you have noticed any problems, please let me know, and you could put following tunables in /boot/loader.conf: hw.tsc_cputimer_enable=0 to disable TSC cputimer. sysctl kern.cputimer will

SO_REUSEPORT and accept(2) performance

Hi all, The brief summary (as of 9a1272865c3cb6e079e1554031dcc712a881598b): nonblocking accept(2)+kqueue(2) w/ SO_REUSEPORT, we are doing 335Kconns/s nonblocking accept(2)+kqueue(2) w/o SO_REUSEPORT, we are doing 100Kconns/s On DragonFly, in addition to better load balance, SO_REUSEPORT also

Re: Is there any policy routing tool in dragonflybsd

On Sat, Aug 17, 2013 at 7:00 PM, jinge altman87...@gmail.com wrote: Hi everyone! Is there any policy routing tool in dragonflybsd like the setfib in FreeBSD? We don't have multiple routing table (mutiple FIB) support yet. Best Regards, sephe -- Tomorrow Will Never Die

Re: dragonflybsd router

On Tue, Aug 20, 2013 at 4:37 PM, k simon chio1...@gmail.com wrote: Hi,list: I have some linux router/NAT box, each can serve 500-600Kpps traffic. Last week, I observed the ddos attack, it report ipt_netflow sendbuffer overlimit and cpu usage high. So I took a look at BSD family. I've

Re: dragonflybsd router

to test dfly BSD with real traffic when the next version released When multiple routing table is done, I will post to users@. Best Regards, sephe . Best Regards, Simon 在 2013-8-20，下午6:43， Sepherosa Ziehau 写道： On Tue, Aug 20, 2013 at 4:37 PM, k simon chio1...@gmail.com wrote: Hi

nginx SO_REUSEPORT patch

Hi all, The nginx httperf result w/ SO_REUSEPORT: 49852 reqs/s There are still 35%~40% idle time on each hyperthread (it probably could do more, however, I can't find more boxes to run httperf on :P). The nginx httperf result w/o SO_REUSEPORT: 37386 reqs/s SO_REUSEPORT gives ~33% performance

Re: DF as KVM Guest Clock Skew - any experiences?

On Tue, Oct 22, 2013 at 10:36 AM, Chris Turner c.tur...@199technologies.com wrote: Hello - Trying to run a 64bit DF guest on F18, am getting some odd clock jitter both with and without dntpd running from the guest, and needless to say, 'make' doesn't like this. What's the output of

Re: what's mean waitport/waitmsg

net.inet.tcp.recvspace=32768 net.inet.tcp.recvbuf_auto=0 net.inet.tcp.sendbuf_auto=0 net.inet.tcp.tso=0 kern.metadelay=3 kern.dirdelay=4 kern.filedelay=5 vfs.write_behind=0 vfs.ufs.dirhash_maxmem=16777216 Regards Simon On 2013年10月24日, at 下午12:12, Sepherosa Ziehau sepher...@gmail.com wrote

Re: BigData (Hadoop/NoSQL)

We implemented RFC6675 and it is enabled by default (IMHO, all modern TCP/IP stacks are SACK enable). You could read this http://tools.ietf.org/html/rfc6937 section.6, to get the general information. Best Regards, sephe On Wed, Oct 30, 2013 at 10:38 AM, Muhammad Nuzaihan Bin Kamal Luddin

Re: (i386) 3.6.0 ifconfig bge0 gives Fatal trap 19.

On Sun, Dec 1, 2013 at 9:58 PM, Makoto Fujiwara mak...@ki.nu wrote: I've another DL320/G5p, very much similar configuration with Dual core CPU. The first one was with 1GB memory. This second one has 8GB memory. 1. Have USB bootimage written to USB memory stick. 232361930 Nov 27 10:21

ACPICA 20131218 update available for testing

Hi all, ACPICA 20131218 is available for testing (master): http://gitweb.dragonflybsd.org/~swildner/dragonfly.git/shortlog/refs/heads/acpi-20131218 Swildner, jh32 an I have done some preliminary test. All seem fine; no big surprise :) Please help testing. Thanks. Best Regards, sephe --

Better ACPI C-state support

Hi all, I have made good progress to improve ACPI C-state support on Intel CPUs (x86_64). All necessary bits are pushed to master. Please test. For many Intel CPU based boxes, which only had C1, now more C-states should be available. And for CPUs' which have ARAT, i8254 is no longer required

MWAIT C-state hints

Hi all, I have pushed some changes to enable idle wait's mwait C-state hint auto tuning. It will allow you enter much deeper C-state, even when ACPI does not map its deepest C-state to the deepest cpu specific C-state (mwait hint). e.g. on my laptop, w/o power cable, ACPI C3 is mapped to cpu

Re: HEADSUP: New ACPI in master

On Thu, Feb 6, 2014 at 11:15 AM, Matthew Dillon dil...@apollo.backplane.com wrote: :Hi everybody, : :I've upgraded ACPI in master on Monday. : :If you are seeing ACPI related error messages on the console or notice any :other weird issues that might be related, please make sure that your

sysctl hw.cpu_mwait_{idle,spin} is deprecated on x86_64

Hi all, hw.cpu_mwait_{idle,spin} sysctls are now read-only on x86_64. machdep.mwait.CX.idle and machdep.mwait.CX.spin should be used instead. The possible values for the above two sysctls are listed in the output of sysctl machdep.mwait.CX.supported. For machdep.mwait.CX.idle, AUTO and

Re: crash of unknown cause

On Wed, Feb 5, 2014 at 11:05 PM, Pierre Abbat p...@leaf.dragonflybsd.org wrote: On Wednesday, February 05, 2014 22:47:21 Sepherosa Ziehau wrote: Hi, Have you enabled RED or RIO in the altq configuration? I saw a problem in the related code. If that's a kernel config, no. I'm using

HEADS UP: p4tcc is replaced by clockmod in master

Hi all, I have replaced i386 only and buggy p4tcc w/ clockmod(4). Kernel configuration: option CPU_ENABLE_TCC is no longer supported. Instead you should use: device clockmod. The new clock modulation domain is controlled through sysctls: machdep.clockmod_domX.select

Re: 10G network netperf performance (ix)

On Fri, Mar 14, 2014 at 6:10 PM, Dongsheng Song dongsheng.s...@gmail.com wrote: On Thu, Mar 13, 2014 at 9:26 PM, Sepherosa Ziehau sepher...@gmail.com wrote: Hi all, Following stats are for folks interested in DragonFly's TCP netperf performance on 10G network (as of 9f1b012): Testing

Re: 10G network netperf performance (ix)

On Fri, Mar 14, 2014 at 8:02 PM, Sepherosa Ziehau sepher...@gmail.com wrote: Thanks, could you post TCP_RR data ? I am not sure whether TCP_RR is really useful, since each process is working on one socket. However, I have some statistics for tools/tools/netrate/accept_connect

IPv6 patch to try

Hi all, Please test the following IPv4 patch: http://leaf.dragonflybsd.org/~sephe/in6_pcbbind.diff Need IPv6 TCP based server start test and IPv6 TCP client (connect) tests. Best Regards, sephe -- Tomorrow Will Never Die

Re: other tests with dragonfly

On Sun, May 4, 2014 at 3:26 PM, Wojciech Puchar woj...@wojtek.tensor.gdynia.pl wrote: For what it's worth, Sepherosa Ziehau has been putting a lot of work into reducing CPU use on high-bandwidth connections: http://lists.dragonflybsd.org/pipermail/users/2014-March/128240.html if i understand

Re: other tests with dragonfly

On Sun, May 4, 2014 at 4:06 PM, Wojciech Puchar woj...@wojtek.tensor.gdynia.pl wrote: do you have some tests on good NIC how much CPU is used not for connection but simply for large TCP transfers? On 82599, I am seeing 75% ~ 85% idle time on i7-3770 on each HT when TCP sending at 9420Mbps (128

HEADS UP: x86_64, GSI target CPU auto selection is enabled by default again

Hi all, I enabled GSI target CPU auto selection by default again on x86_64. If you noticed any problem, please let me know. Best Regards, sephe -- Tomorrow Will Never Die

Re: lots of bad checksum on igb

On Fri, May 23, 2014 at 12:36 AM, k simon chio1...@gmail.com wrote: Hi,Lists, I noticed that Pyun have fixed some bugs about TSO.CHECKSUM offload issues on FreeBSD recently. Does the same issues happend on Dfly ? I don't aware we have any TSO and TX hardware CSUM issues. And I don't think

Re: ipfw2 for dragonflybsd

For static rules, ipfw2 in the base is already lockless MPSAFE. I want to change the states lockless MPSAFE (it was not doable back to the time I made ipfw2 static rules lockless MPSAFE; but now since we dispatch udp outputs to the proper netisr, it is doable). On Mon, Nov 17, 2014 at 6:13 PM,

Re: ipfw2 for dragonflybsd

, Nov 19, 2014 at 9:26 AM, bycn82 byc...@gmail.com wrote: why not share more information, for example write a internal document for developers like NetBSD in order to help newcomers. https://www.netbsd.org/docs/internals/en/index.html On Tue, Nov 18, 2014 at 6:45 AM, Sepherosa Ziehau sepher

Re: ipfw2 for dragonflybsd

its bwteen at most two CPUs (two netisrs).\ Best Regards, sephe Anyway , next will be lockless for keep-state. maybe this weekend :) On Fri, Dec 5, 2014 at 3:11 PM, Sepherosa Ziehau sepher...@gmail.com wrote: On Fri, Dec 5, 2014 at 2:38 AM, Matthew Dillon dil...@apollo.backplane.com

Re: Acer c720 machdep.mwait.CX.[idle, spin]

On Tue, Dec 9, 2014 at 4:51 AM, Matthew Dillon dil...@apollo.backplane.com wrote: : Also suggested in /etc/sysctl.conf: : : hw.cpu_mwait_halt=0x44 : hw.cpu_mwait_spin=0x11 : : Now these sysctls are read-only. Which C-states I must use with : machdep.mwait.CX.[idle, spin]? : :

Re: ipfw2 for dragonflybsd

was there, but any recommendation about the testing in order to know the performance different? On Fri, Dec 5, 2014 at 5:25 PM, Sepherosa Ziehau sepher...@gmail.com wrote: On Fri, Dec 5, 2014 at 3:39 PM, bycn82 byc...@gmail.com wrote: It will be good to have a try on this lock-less NAT

HEADS UP: V4-mapped address support is removed

Hi all, V4-mapped address is no longer supported in the master, since 727ccde8cce813911d885b7f6ed749dcea68a886. Both kernel and world need rebuilding, else netstat, systat and sockstat may give you weird results. I have removed the inpcb.inp_vflag, INP_IPV4 and INP_IPV6 which may cause some

HEADS UP: TCP path mtu discovery is enabled by default

Hi all, TCP path mtu discovery is enabled by default on master. If you notice problem please let me know. You could disable it by: sysctl net.inet.tcp.path_mtu_discovery=0 Thanks, sephe -- Tomorrow Will Never Die

Re: Cannot install xorg on dfly-x86_64-4.0.1_REL

On Sat, Dec 27, 2014 at 4:04 PM, Aero 9000 mbg11...@gmail.com wrote: Hi Folks, I'm not sure if this is the correct mailing list for me to post this to, but if not just tell me and I'll be out of your hair. Anyways, I've known of DragonFly BSD for a couple of years now, just never got round to

SOL_SOCKET/SO_CPUINT socket option

Hi all, SOL_SOCKET/SO_CPUHINT socket option (currently read-only). This option gives you the hint about to which CPU the userland process could be bound to (-1 means no hint). It's mainly for TCP/UDP server userland application. For TCP listen socket, it only makes sense for TCP listen socket

Re: SATA chips supported

On Sat, Jan 17, 2015 at 11:34 PM, Jasse Jansson ja...@yberwaffe.com wrote: I just wonder if the SATA chips are supported in DFly Marvell 88se9120 Asmedia 1061 I am pretty sure that asmedia chips are supported by ahci(4). 9120 also attached to ahci(4), last time I tested. Best Regards,

Re: Performance Counter Tools?

to the 4.0 RELEASE head? I think it could be cherry-picked to 4.0.x branch. Best Regards, sephe Thanks, Alex On Thu, Apr 23, 2015 at 9:28 PM, Sepherosa Ziehau sepher...@gmail.com wrote: On Fri, Apr 24, 2015 at 2:32 AM, Alex Merritt merritt.a...@gmail.com wrote: Hello, Can anyone

Re: DragonFly power management stuffs

On Thu, Jul 23, 2015 at 5:27 PM, Bob Bishop r...@gid.co.uk wrote: Hi, On 23 Jul 2015, at 02:22, Sepherosa Ziehau sepher...@gmail.com wrote: First of all, I'd like to introduce the major power management stuffs currently available on DragonFly. - ACPI P-state. It has the proper CPU power

DragonFly power management stuffs

First of all, I'd like to introduce the major power management stuffs currently available on DragonFly. - ACPI P-state. It has the proper CPU power domain support. - ACPI C-state. Unlike other BSDs, on relatively recent Intel CPUs (the oldest Intel CPU I tested is Sandy Bridge), we don't use

Re: Storing hundreds of millions of files in HAMMER (1 or 2)

Maybe just use large file, and sub-index the chunks of a large file and add open/read/write/lseek/close like APIs for users. You will have more control than using database. On Wed, Jul 15, 2015 at 11:58 PM, Michael Neumann mneum...@ntecs.de wrote: Hi, Lets say I want to store 100 million

Re: IPv6 DNS config daemon

It could be useful. Is it available somewhere? So that we could take a look at it. Thanks, sephe On Fri, Oct 23, 2015 at 10:57 AM, Charles Musser wrote: > While experimenting with IPv6 SLAAC, I wrote a daemon that configures > a host's DNS resolver using the scheme

Re: Dell Inspiron 15 5000 series

Put the following lines in /boot/loader.conf: coretemp0.disabled="1" coretemp1.disabled="1" coretemp2.disabled="1" coretemp3.disabled="1" On Sun, Nov 1, 2015 at 3:17 AM, Wojciech Puchar wrote: > tried to run dragonfly on it. > > It boots fine and hardware is detected BUT i'm

Re: getting cpu info w/o parsing /var/run/dmesg.boot

Those information could be exported through sysctl easily. Anyone want to take a simple kernel hacking task to complete it? I believe them to be pretty useful. Thanks, sephe On Fri, Oct 9, 2015 at 11:47 PM, Hleb Valoshka <375...@gmail.com> wrote: > On 10/9/15, Zachary Crownover

HEADS UP: master kernel upgrade will be required

Hi all, For master users only. If you caught the following commit or above: db38e955e08f32a306e5ce7e118afc6fdd05966e Please at least upgrade your kernel to: ac7b8e28062207a6d6db1b0bd95207755749fa7c The problematic commit will cause panic on tcp servers. Sorry for the breakage. Thanks, sephe

Re: Call for test (em/emx/igb update)

Hmm, please re-pull that branch. I added some debug prints, so that we could narrow it down. Thanks, sephe On Thu, Dec 17, 2015 at 11:01 AM, Matthew Dillon wrote: > emx doesn't work for me either, same error. > > emx0: port 0xf080-0xf09f

Re: Call for test (em/emx/igb update)

I am closing that branch after adding some debug prints. It seems that the EEPROM could not be read on I217/I218. But if you have time, please test the HEAD of that branch, and post the debug messages; it will help the next em update attempt :). I'd thank Matt very much for the tedious

Call for test (em/emx/igb update)

Please test this branch with your Intel NICs (em/emx/igb): http://gitweb.dragonflybsd.org/~sephe/dragonfly.git/shortlog/refs/heads/em752igb243 I have tested 82571, 82574 and 82573 (em/emx); 82575, 82576, 82580 and i350, so it should be safe to test this branch :). Thanks, sephe -- Tomorrow

Re: re0 drivers (was tips for writing img to usb thumb drive)

On Sun, Jan 3, 2016 at 5:33 PM, jungle Boogie <jungleboog...@gmail.com> wrote: > On 2 January 2016 at 23:59, Sepherosa Ziehau <sepher...@gmail.com> wrote: >> I have pushed some changes to the re's PHY initialization, you could >> try latest usb image; hope it works for

Re: Call for test (em/emx/igb update)

On Fri, Dec 18, 2015 at 3:57 PM, Sepherosa Ziehau <sepher...@gmail.com> wrote: > I am closing that branch after adding some debug prints. It seems > that the EEPROM could not be read on I217/I218. But if you have time, > please test the HEAD of that branch, and post the

HEAD UP: hammer mount is forced noatime by default (master)

Hi all, If you catch this commit: 4aaf46d1d57bf0c18aa78f35628f8506dc637afb Hammer mount is then forced to noatime by default. If you already has noatime in /etc/fstab for hammer, that's fine. But if you do want atime, you have to set vfs.hammer.noatime=0 in /boot/loader.conf and reboot.

Re: IPv6 DNS config daemon

On Fri, Oct 23, 2015 at 2:23 PM, Sepherosa Ziehau <sepher...@gmail.com> wrote: > It could be useful. Is it available somewhere? So that we could take > a look at it. > > Thanks, > sephe > > On Fri, Oct 23, 2015 at 10:57 AM, Charles Musser <cmus...@sonic.net&

accept(2)/close(2) performance on DragonFly-1ae4ae5

The network configuration and various settings (tunables/sysctls) are same as: http://lists.dragonflybsd.org/pipermail/users/2016-May/249647.html The accept(2)/close(2) performance is measured w/:

Re: HandBrake CLI tools on Dragonfly

On Mon, May 30, 2016 at 9:18 PM, PeerCorps Trust Fund wrote: > Hi all, > > The multimedia tool HandBrake obviously does not build on Dragonfly > > http://muscles.dragonflybsd.org/latest-failures/logs/errors/handbrake-0.10.5.log Looks like you need to copy the

Re: nginx-1.9.12 performance on DragonFly-g680c4d.

On Fri, May 27, 2016 at 5:45 AM, Carsten Mattner <carstenmatt...@gmail.com> wrote: > On Thu, May 26, 2016 at 4:08 PM, Sepherosa Ziehau wrote: > >> And for 16KB object, we now max out 2x10Ge (19.2Gbps, was 16.8Gbps >> before 680c4dd) >> >> This will serve

Patch for I219

Hi all, Please test following patch on you Intel NICs, it mainly affects em/emx: https://www.dragonflybsd.org/~sephe/0001-ig_hal-em-emx-Add-I219-Skylake-support.patch Thanks, sephe -- Tomorrow Will Never Die

HEADS UP: world reneeds rebuilding 6de344baaa

For master users only. If you caught 6de344baaa1923ba65e83c698fc34ae686de07dc, both kernel and world would need rebuilding. The stuffs like netstat in the old world will _not_ work properly w/ new kernel. Thanks, sephe -- Tomorrow Will Never Die

Re: Some artifacts - bad screen on Lumina-DE and Firefox

On Fri, Jun 17, 2016 at 11:18 AM, Dmitry Postolov wrote: > Hi to all! Sorry for my bad English... > > I am faced some artifacts in Lumina-DE and Firefox when scrolling > the screen of web-page by mouse wheel or arrows on the keyboard, > the text/image in rare instances is

hammer streaming seems stop working for a while

Hi all, I just noticed today, my hammer back mechanism seems stop working for a while. The output of mount on the backup machine: ROOT on / (hammer, local, noatime) devfs on /dev (devfs, local) /dev/serno/3PT023B3.s1a on /boot (ufs, local, noatime) REPOS on /repos (hammer, local, noatime)

HEADS UP: rebuild world is needed

Hi all, It only affects master users. If you get 4a43469a10cef8c17553c342aab9d73611ea7bc8, then world needs rebuilding, else things like netstat will not work properly. Thanks, sephe -- Tomorrow Will Never Die

Re: dmesg troubleshooting

On Tue, Apr 5, 2016 at 11:09 AM, Predrag Punosevac <punoseva...@gmail.com> wrote: > Sepherosa Ziehau <sepher...@gmail.com> wrote: > >> Please post the output of 'vmstat -iv' > > dfly# vmstat -iv > interruptto

Re: Study of nginx-1.9.12 performance/latency on DragonFlyBSD-g67a73.

On Tue, May 10, 2016 at 1:40 AM, Matthew Dillon wrote: > Very interesting results. I'm not surprised that lowering the poll rate > improves the results, that's an artifact of scheduling larger batches (as > long as the ring buffer doesn't fill up completely). It's true for

Study of nginx-1.9.12 performance/latency on DragonFlyBSD-g67a73.

Study of nginx-1.9.12 performance/latency on DragonFlyBSD-g67a73. The performance and latency is measured using a modified version of wrk: https://github.com/sepherosa/wrk.git (sephe/wrk branch). It mainly adds requests/connection setting and avoids several unnecessary syscalls. Hardware

Re: 2 Questions!

On Sun, Jul 31, 2016 at 3:46 AM, Mohammad BadieZadegan wrote: > Hi, > Still I'm waiting for your good answers. > Thanks. > > > On Wed, Jul 27, 2016 at 10:08 AM, Mohammad BadieZadegan > wrote: >> >> Hi everybody, >> I have 2 questions about DragonflyBSD,

Are 4.8 packages are ready?

Got following errors on 4.7 systems: xanadu1:~# pkg update Updating Avalon repository catalogue... Repository Avalon has a wrong packagesite, need to re-create database meta.txz : 100%264 B 0.3kB/s00:01 packagesite.txz : 100%5

Re: Random server crashes every few weeks (smp_invltlb: endless loop […] retrysmp_invltlb: ipi sent)

On Wed, Jul 6, 2016 at 6:17 AM, Stefan Unterweger <232.20...@chiffre.aleturo.com> wrote: > Sorry for the long delay. Other, more urgent projects kept me so busy > that I didn’t manage to catch up with this thread. > > > * Matthew Dillon on Fri, May 27, 2016 at 10:08:06AM -0700: >> Virtio (for

Re: Kernel panic on boot at dm_crypt

I am compiling right > now, as soon as I manage to push it to the server, I‘ll report back. > > | Author: Sepherosa Ziehau <se...@dragonflybsd.org> 2016-07-06 10:17:21 > | Committer: Sepherosa Ziehau <se...@dragonflybsd.org> 2016-07-06 10:19:45 > | Parent: 695396069af2

Re: \build directory full?

On Thu, Jul 7, 2016 at 6:27 AM, Andrew Slaughter wrote: > I just tried to update to the current master, and the build failed with > messages about having no space left on /usr/obj. > > df -h shows me the following (apologies for horrible formatting): > > Filesystem

Re: Kernel panic on boot at dm_crypt

On Thu, Jul 7, 2016 at 10:57 AM, Sepherosa Ziehau <sepher...@gmail.com> wrote: > On Wed, Jul 6, 2016 at 5:54 AM, Stefan Unterweger > <232.20...@chiffre.aleturo.com> wrote: >> Yet another problem on the same virtualised host as in the other thread >> a f

Re: My testing of Gnome3 on Sony Laptop

On Thu, Jul 7, 2016 at 2:31 PM, Bernard Mentink wrote: > Hi All, > > I would just like to report my results of trying out DFly on a 2009 Sony > Vaio VGN-FW laptop, especially as I had good results running Gnome3 on a > Desktop PC. > > After much testing this is what I found as

Re: Kernel panic on boot at dm_crypt

erence panic. However, I don't know why the I/O happens simultaneously w/ the init. > > 2016-07-07 12:16 GMT+09:00 Sepherosa Ziehau <sepher...@gmail.com>: >> On Thu, Jul 7, 2016 at 10:57 AM, Sepherosa Ziehau <sepher...@gmail.com> >> wrote: >>> On Wed, Ju

Re: powerdxx

On Mon, Aug 1, 2016 at 5:37 PM, PeerCorps Trust Fund wrote: > Does anyone have any experience using powerdxx under Dragonfly? > > https://www.freshports.org/sysutils/powerdxx/ > > It appeared in dports not too long ago and seems to be quite an improvement > over the

Re: Jails vs Vkernel

On Sat, Jan 21, 2017 at 1:30 PM, wrote: > Greetings. So, a question / request for feedback about isolation... > > Setting up a DF machine where isolation of the various public facing services > is desired. After reading about Vkernels, it seems like a perfect solution > for

Re: CFT: Realtek network driver update

On Wed, Feb 22, 2017 at 12:13 AM, Tim Darby <t+df...@timdarby.net> wrote: > On Fri, Feb 17, 2017 at 11:46 AM, Tim Darby <t+df...@timdarby.net> wrote: >> >> On Thu, Feb 16, 2017 at 7:16 AM, Sepherosa Ziehau <sepher...@gmail.com> >> wrote: >>>

Re: DragonFly 4.6 network not working when vm guest

What's the NIC? And what's the emulated NIC mode in the vmware? On Sat, Feb 25, 2017 at 9:11 PM, Tomohiro Kusumi wrote: > This reproduces with DragonFly 4.4 as well on VMware. > > wget of dfly-x86_64-4.6.1_REL.iso stops at around after it has > received the first

CFT: Realtek network driver update

For master users only. Please test the following patch: https://leaf.dragonflybsd.org/~sephe/re_193.diff Thanks, sephe -- Tomorrow Will Never Die

Re: CFT: Realtek network driver update

On Fri, Feb 24, 2017 at 5:34 PM, Daniel Bilik <d...@neosystem.org> wrote: > On Fri, 24 Feb 2017 09:43:31 +0800 > Sepherosa Ziehau <sepher...@gmail.com> wrote: > >>> Second problem is more tough (or weird, I'd say), and AFAICT is >>> related to txcsum.

Re: DragonFly 4.6 network not working when vm guest

On Sun, Feb 26, 2017 at 8:56 PM, Tomohiro Kusumi wrote: > Now I started to see this on FreeBSD as well (which I didn't for the > last few days). > Linux is the only one that's properly working. > So it might be just something else. Did vmware bridge mode emulation work

DSA keys are disabled by default on master

Hi all, DSA keys are disabled by default on master, so make sure you have some non-DSA pub-keys deployed before its too late on you ssh servers. For ssh to use DSA keys, put following line into your .ssh/config: PubkeyAcceptedKeyTypes=+ssh-dss Thanks, sephe -- Tomorrow Will Never Die

Call for test [AMD erratum 793]

AMD cpu users please help testing following patch: https://www.dragonflybsd.org/~sephe/0001-x86_64-Work-around-AMD-erratum-793-for-family-16h-mo.patch -- Tomorrow Will Never Die

Re: Monitoring CPU time

Hi, The bump upon each statclock is: ((cur_systimer - prev_systimer) * systimer_freq) >> 32 systimer_freq can be extracted from following sysctl in userspace: sysctl kern.cputimer.freq statclock is called at stathz frequency. Thanks, sephe On Mon, Sep 19, 2016 at 12:46 AM, Stuart Nelson

Re: extreme network latency

On Thu, Oct 20, 2016 at 2:22 AM, Richard Nyberg wrote: >> For better performance, you can put the if_emx_load="YES" in >> loader.conf. However, em0 will become emx0 after loading that module, >> so make sure that your rc.conf and pf.conf are also updated ;) > > It turned

Re: extreme network latency

On Tue, Oct 18, 2016 at 4:17 AM, Richard Nyberg wrote: > Yes, that was it. Many thanks! > > Should I just use polling, which works fine, or is there something one > can do about the interrupt issue? Heh, I'd say avoid re :). Try put the following tunable:

Re: extreme network latency

On Wed, Oct 19, 2016 at 2:19 AM, Richard Nyberg <rnyb...@murmeldjur.se> wrote: > Hi! Yes it's quite recent hardware. > > On 18 October 2016 at 04:20, Sepherosa Ziehau <sepher...@gmail.com> wrote: >> >> Heh, I'd say avoid re :). > > I might not have made t

CFT: new re(4)

Hi all, I have made an updated version of re(4), which leverages Realtek driver's chip/PHY reset/initialization code. I hope it can resolve all kinds of weirdness we encountered on this chip so far. Please help testing this branch:

Re: HAMMER, disk mirroring and secure NFS

On Sun, Mar 26, 2017 at 9:02 PM, Jasse Jansson wrote: > > > On 2017-03-26 14:41, Tomohiro Kusumi wrote: >> >> You can either directly copy between fses, or you can dump the binary >> from one fs to stdout (and redirect it to a file) and later restore. >> Both are technically

Summary of my trip to AsiaBSDCon 2017

Hi all, It's a quite nice conference. And Tokyo is a great place for sightseeing. I highly recommend DragonFly folks to join next year's AsiaBSDCon. Gave two talks this time, one is for DragonFly's network stack: https://leaf.dragonflybsd.org/~sephe/AsiaBSDCon%20-%20Dfly.pdf

Re: Summary of my trip to AsiaBSDCon 2017

On Fri, Mar 17, 2017 at 4:19 AM, Carsten Mattner <carstenmatt...@gmail.com> wrote: > On Wed, Mar 15, 2017 at 1:55 PM, Sepherosa Ziehau <sepher...@gmail.com> wrote: > >> Met a group of nice FreeBSD folks; talked a bit about the recent >> network performance compari

Re: dmesg troubleshooting

com> wrote: > Sepherosa Ziehau <sepher...@gmail.com> wrote: > >> Can you post the dmesg and vmstat -iv? I think for ahci controller, >> MSI will be used by default. > > dfly# vmstat -iv > interrupttotal rate > irq1 0: atk

Re: Network performance comparison as of today.

, Mar 3, 2017 at 3:44 PM, Sepherosa Ziehau <sepher...@gmail.com> wrote: > Hi all, > > Since so many folks are interested in the performance comparison, I > just did one network related comparison here: > https://leaf.dragonflybsd.org/~sephe/perf_cmp.pdf > > The

Network performance comparison as of today.

Hi all, Since so many folks are interested in the performance comparison, I just did one network related comparison here: https://leaf.dragonflybsd.org/~sephe/perf_cmp.pdf The intention is _not_ to troll, but to identify gaps, and what we can do to keep improving DragonFlyBSD. According to the

Fwd: BSDTW 2017 still don't have your talk proposal!

BSDTW! -- Forwarded message -- From: Li-Wen Hsu Date: Sun, Jul 9, 2017 at 8:56 AM Subject: BSDTW 2017 still don't have your talk proposal! Howdy, We would like to announce TWBSD, coming up on 11-12 November 2017 in Taipei. You can find the details at:

Re: DragonFly and Hardware P states in Skylake

Hi, We by default turn on the HW coordination. For most of the modern Intel CPUs, the P-state setting, hw.acpi.cpu.px_domX, only puts the upper limit of the CPUs frequency; e.g. CPU may run @1000Mhz even if the P-state is set to 2000Mhz, when the system is idle. As far as I tested, most of the

Re: ezjail

On Thu, May 11, 2017 at 12:42 AM, Matthias wrote: > Hi, > > found that the crash was introduced with commit related to > multi-threaded UDP traffic handling. > > When reverting changes done in that commit in latest CURRENT the crash > does not occur anymore and ezjail works

Re: newbie questions

On Wed, May 3, 2017 at 4:25 PM, bill . wrote: > Hi all, > > > I am a long-time Linux user with a little BSD experience. No guru, just a > guy who tries to steer away from the megalo corporations. I would like to > try Dragonfly but have had some issues in the past. Using Linux,

Re: [Bug #3032] IPFW3: memory leakage? objcache(xxx): Exhausted!

On Fri, May 5, 2017 at 9:43 AM, Aaron LI <aaronly...@outlook.com> wrote: > > On Fri, May 05, 2017 at 01:16, Sepherosa Ziehau <sepher...@gmail.com> wrote: > >> On Thu, May 4, 2017 at 7:44 PM, Aaron LI <aaronly...@outlook.com> wrote: >>> Dear Bill, >&g

Re: [DragonFlyBSD - Bug #3032] (New) IPFW3: memory leakage? objcache(xxx): Exhausted!

On Thu, May 4, 2017 at 7:44 PM, Aaron LI wrote: > Dear Bill, > > I'm running into problems with the IPFW3, which seems to cause memory > leakages and lead to the "objcache(xxx): Exhausted!" warning, and > finally the system became unresponsive and need reset. > > On the

BSDTW 2017 registration opened

In case you are interested: https://bsdtw.org/ The site looks nice :) Thanks, sephe -- Tomorrow Will Never Die

Network improvement 4.8 -> 5.0 for short-lived HTTP/1.1 workload.

In this release cycle, several items are committed to improve performance and reduce/stablize latency. 30K concurrent connections, 1 request/connection, 1KB web object. Server have 24 HT. Baseline (32 nginx workers w/ 16 netisrs): performance 215907.25tps, latency-avg 33.11ms, latency-stdev

  1   2   >