Search Results

Search found 18521 results on 741 pages for 'tcp window scaling'.

Page 6/741 | < Previous Page | 2 3 4 5 6 7 8 9 10 11 12 13  | Next Page >

  • Exploring TCP throughput with DTrace

    - by user12820842
    One key measure to use when assessing TCP throughput is assessing the amount of unacknowledged data in the pipe. This is sometimes termed the Bandwidth Delay Product (BDP) (note that BDP is often used more generally as the product of the link capacity and the end-to-end delay). In DTrace terms, the amount of unacknowledged data in bytes for the connection is the different between the next sequence number to send and the lowest unacknoweldged sequence number (tcps_snxt - tcps_suna). According to the theory, when the number of unacknowledged bytes for the connection is less than the receive window of the peer, the path bandwidth is the limiting factor for throughput. In other words, if we can fill the pipe without the peer TCP complaining (by virtue of its window size reaching 0), we are purely bandwidth-limited. If the peer's receive window is too small however, the sending TCP has to wait for acknowledgements before it can send more data. In this case the round-trip time (RTT) limits throughput. In such cases the effective throughput limit is the window size divided by the RTT, e.g. if the window size is 64K and the RTT is 0.5sec, the throughput is 128K/s. So a neat way to visually determine if the receive window of clients may be too small should be to compare the distribution of BDP values for the server versus the client's advertised receive window. If the BDP distribution overlaps the send window distribution such that it is to the right (or lower down in DTrace since quantizations are displayed vertically), it indicates that the amount of unacknowledged data regularly exceeds the client's receive window, so that it is possible that the sender may have more data to send but is blocked by a zero-window on the client side. In the following example, we compare the distribution of BDP values to the receive window advertised by the receiver (10.175.96.92) for a large file download via http. # dtrace -s tcp_tput.d ^C BDP(bytes) 10.175.96.92 80 value ------------- Distribution ------------- count -1 | 0 0 | 6 1 | 0 2 | 0 4 | 0 8 | 0 16 | 0 32 | 0 64 | 0 128 | 0 256 | 3 512 | 0 1024 | 0 2048 | 9 4096 | 14 8192 | 27 16384 | 67 32768 |@@ 1464 65536 |@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@ 32396 131072 | 0 SWND(bytes) 10.175.96.92 80 value ------------- Distribution ------------- count 16384 | 0 32768 |@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@ 17067 65536 | 0 Here we have a puzzle. We can see that the receiver's advertised window is in the 32768-65535 range, while the amount of unacknowledged data in the pipe is largely in the 65536-131071 range. What's going on here? Surely in a case like this we should see zero-window events, since the amount of data in the pipe regularly exceeds the window size of the receiver. We can see that we don't see any zero-window events since the SWND distribution displays no 0 values - it stays within the 32768-65535 range. The explanation is straightforward enough. TCP Window scaling is in operation for this connection - the Window Scale TCP option is used on connection setup to allow a connection to advertise (and have advertised to it) a window greater than 65536 bytes. In this case the scaling shift is 1, so this explains why the SWND values are clustered in the 32768-65535 range rather than the 65536-131071 range - the SWND value needs to be multiplied by two since the reciever is also scaling its window by a shift factor of 1. Here's the simple script that compares BDP and SWND distributions, fixed to take account of window scaling. #!/usr/sbin/dtrace -s #pragma D option quiet tcp:::send / (args[4]-tcp_flags & (TH_SYN|TH_RST|TH_FIN)) == 0 / { @bdp["BDP(bytes)", args[2]-ip_daddr, args[4]-tcp_sport] = quantize(args[3]-tcps_snxt - args[3]-tcps_suna); } tcp:::receive / (args[4]-tcp_flags & (TH_SYN|TH_RST|TH_FIN)) == 0 / { @swnd["SWND(bytes)", args[2]-ip_saddr, args[4]-tcp_dport] = quantize((args[4]-tcp_window)*(1 tcps_snd_ws)); } And here's the fixed output. # dtrace -s tcp_tput_scaled.d ^C BDP(bytes) 10.175.96.92 80 value ------------- Distribution ------------- count -1 | 0 0 | 39 1 | 0 2 | 0 4 | 0 8 | 0 16 | 0 32 | 0 64 | 0 128 | 0 256 | 3 512 | 0 1024 | 0 2048 | 4 4096 | 9 8192 | 22 16384 | 37 32768 |@ 99 65536 |@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@ 3858 131072 | 0 SWND(bytes) 10.175.96.92 80 value ------------- Distribution ------------- count 512 | 0 1024 | 1 2048 | 0 4096 | 2 8192 | 4 16384 | 7 32768 | 14 65536 |@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@ 1956 131072 | 0

    Read the article

  • centos iptables, restrict tcp port to specific ips

    - by user788171
    I would like to modify the iptables on my CentOS 5.8 server so that only specific ips can connect to the machine on a specific port. Currently, I have the following in my iptables file: -A RH-Firewall-1-INPUT -m state --state NEW -m tcp -p tcp --dport 5000 -j ACCEPT How would I modify that line if I wanted to allow access for only ips 1.1.1.1 and 1.1.1.2 for instance? (they might not necessarily be sequential ips when I do this for reals).

    Read the article

  • structure of ethernet frame (tcp/udp) [closed]

    - by rtmrtm2
    How is an ethernet-frame structured. is it: |MAC | |_______________| | |IP | | |___________| | |TCP | | |_______| | |HTTP| |__________|____| or the other way around? so in words: is the mac wrapped around the ip wrapped around the tcp wrapped arround the http? can someone post an image of the specific 'wrapping'? thanks in advance. regards, rtmrtm2

    Read the article

  • TCP Proxy with multiple clients?

    - by Daphna
    I am looking for a TCP proxy - a utility that will connect to a port, read a TCP stream, and write it to clients that connect to it. The key point here is that there may be more than one client, and each client should receive a copy of the stream. Preferably windows solution, but Linux can be useful as well.

    Read the article

  • Change TCP wait for ACK timeouts in Win7/WinServer

    - by maseth
    Is there any possibility to change default wait for ACK timeout in TCP network on Windows 7 or Windows Server ? I'm using very slow network ( 1200 bps ) and want to tweak TCP. When using default parameters network stuck on multiple retransmissions . If I'm able to change the ACK timeout and tx window size I think that it would work. On Windows XP it was possible but cant find any document for Win7 and Win Server.

    Read the article

  • How to measure TCP connection time in Linux

    - by Paul Draper
    I want to measure the overhead in creating a TCP connection. I know of many tools like hping and netperf, but they seem oriented at measuring latency. I want to know how long the 3-way handshake takes, and allocating any buffers, etc., and then closing it. So I want to open a real, legitimate TCP connection, and then close it. Are there any tools that will do that and help me measure performance?

    Read the article

  • Scaling VideoView on Android

    - by Mumbles
    I am creating a video player on android and am having trouble resizing the video. I am trying to make it so that when i click the video it goes to full screen covering everything and if i click again it goes back to the smaller size with other buttons on the screen. I have seen another similar problem and an answer from haseman (http://stackoverflow.com/questions/2068242/does-android-support-scaling-video) and I think this would work but am not sure how to achieve this, Any help would be much appreciated

    Read the article

  • General advice about scaling on QGraphicsView/QGraphicsScene

    - by onurozcelik
    Hi, In my project I am using QGraphicsView/QGraphicsScene stuff. On my scene there will be regions that contains 2D graphics. Region count will be limited(Lets say 20) Users can choose to display one or more regions. If user choose to display one region I am going to show one region on scene If user choose to display n regions I am going to show n regions on scene I need a scaling logic to fit n regions on same scene. How can I achieve this?

    Read the article

  • Scaling ratio of two different resolutions

    - by BBDeveloper
    I have two different resolutions, the original one is 567x756 (wXh), the one which I want to display is 768x1024 (wXh). How to find out the scaling ratio for these two resolutions? For example if the font size used in 567x756 resolution is 7 pts then whats the values I should multiply with the font size (7 pts) to display the text in 768x1024 resolution. Can anyone help me in this issue? Its little urgent. Thanks in advance.

    Read the article

  • Socket options for a tcp server with 3G clients & frequent disconnections

    - by Joel
    I have a TCP server, written in java, sending and receiving many short messages, from 500 bytes to 100 KB long. It's a chess game and chat server, to make it simple. The server is running Debian 6. Half of the clients are connecting from 3G networks, and half over standard DSL. A portion of the 3G clients lose connection pretty often. The error I get on the server and on the client socket is Connection reset. I have come across this page at Oracle documentation: socketOpt. I am wondering what I could tune there to lower the number of disconnections from 3G clients. I don't mind about the ping or transfer rate, but just about the TCP disconnections. I am not skilled enough to understand the impact of each setting, but I sort of understood that the TCP window was important, although I don't know exactly how. So I'm asking if anyone here has an idea ? Thanks if you can help.

    Read the article

  • Puzzling TCP performance over 3G / UMTS

    - by lemonsqueeze
    I'm using 3G as my primary internet connection, and TCP over this thing is getting more puzzling every day. For example: Downloading from kernel.org is crazy fast: $wget http://www.kernel.org/pub/linux/kernel/v3.0/linux-3.6.8.tar.bz2 increases to ~500kB/s after a few secs ! Some servers are incredibly slow, for instance www.graphic-pc.com:Same thing, downloading a big file with wget it starts at ~30kB/s for a split second, then collapses to 5-10k or even worse. Web browsing is decent but somewhat unreliable. Randomly, a page will take really long to load or even fail to load, but a reload can succeed almost immediately. Now, by chance i started playing with OpenVPN over UDP on top of the 3G connection, and OMG suddenly everything's extremely fast !Same www.graphic-pc.com now shoots at 100-200kB/s ! What's going on here ??? How come it is so much better with the VPN than without ?? And why does graphic-pc.com crawl when kernel.org flies ?Something to do with my tcp stack (or the server), or some buggy router in between ?? Notes: Setup is laptop running Ubuntu Lucid and a Huawei 3G dongle (So direct pppd connection). I can reproduce this pretty much any time during the day and I'm not moving, so it's clearly not cell environment or internet congestion. (although kernel.org without VPN sometimes does worse in the evening, 60kB or so - but still 500kB with VPN !) For 2) wireshark shows retransmitted packets, dup ack's, even out of order sometimes. I've tried playing with different /proc/sys/net/ipv4 parameters (tcp_rmem, window_scaling, tcp_congestion...) doesn't seem to make a difference. Update: Tried under windows 7 (no VPN) with some interesting results: tcp settings : default tcp_optimizer kernel.org : 10 kB/s 20 kB/s graphic-pc.com: 8 kB/s 70 kB/s ! tcp_optimizer turned on ctcp among other things. Have to check what os graphic-pc.com is running, my bet is linux's tcp_westwood and ms ctcp don't mix well here...

    Read the article

  • TCP dies on a Linux laptop

    - by Roman Cheplyaka
    Once in several days I have the following problem. My laptop (Debian GNU/Linux testing) suddenly becomes unable to work with TCP connections to the internet. The following things continue to work fine: UDP (DNS), ICMP (ping) — I get instant response TCP connections to other machines in the local network (e.g. I can ssh to a neighbour laptop) everything is ok for other machines in my LAN But when I try TCP connections from my laptop, they time out (no response to SYN packets). Here's a typical curl output: % curl -v google.com * About to connect() to google.com port 80 (#0) * Trying 173.194.39.105... * Connection timed out * Trying 173.194.39.110... * Connection timed out * Trying 173.194.39.97... * Connection timed out * Trying 173.194.39.102... * Timeout * Trying 173.194.39.98... * Timeout * Trying 173.194.39.96... * Timeout * Trying 173.194.39.103... * Timeout * Trying 173.194.39.99... * Timeout * Trying 173.194.39.101... * Timeout * Trying 173.194.39.104... * Timeout * Trying 173.194.39.100... * Timeout * Trying 2a00:1450:400d:803::1009... * Failed to connect to 2a00:1450:400d:803::1009: Network is unreachable * Success * couldn't connect to host * Closing connection #0 curl: (7) Failed to connect to 2a00:1450:400d:803::1009: Network is unreachable Restarting the connection and/or reloading the network card kernel module doesn't help. The only thing that helps is reboot. Clearly something is wrong with my system (everything else works fine), but I have no idea what exactly. I don't know how to reproduce this, but as I said, it happens every several days. My setup is a wireless router that is connected to the ISP via PPPoE. Any advice?

    Read the article

  • How to use different scaling for every monitor?

    - by Mikael Koskinen
    One of the new features in Windows 8.1 is the new "Desktop display scaling", which allows user to configure scaling per monitor. I've been trying to get this working in preview but with no success. If I configure the scaling, it always affects all of my monitors. I have two monitors, the main one with a higher resolution and the secondary with "normal" resolution. The secondary monitor is used in portrait mode. I would like to configure the main monitor's scaling as the text is currently too small. Here's how things look at the configuration screen by default: Now if I adjust the scaling, click apply and do relogin, everything is bigger. On both of my monitors. I haven't clicked the "Let me choose one scaling level for all my displays", but still the slider seems to affect both of them. If I check the "Let me choose one scaling level", the UI changes to look similar to what we have in Windows 8: Still the problem persists. The scaling is applied to both of my monitors. So, it doesn't matter if I check the box or not, the scaling is always applied to all the displays. Any idea how I could get this to work in preview? I've read some comments which seem to indicate that this should work in preview, though Paul Thurrott mentioned at his Winsupersite article that he either didn't get this to work.

    Read the article

  • Window management shortcuts?

    - by pwnguin
    I've got a single massive monitor at home, and I've decided to mimic the Windows 7 window tiling shortcuts. I found a few guides online using wmctrl, and it's going well, save one thing: maximized windows don't respond to it. gconftool-2 --type string --set /apps/metacity/keybinding_commands/command_1 "wmctrl -r :ACTIVE: -e 0, 0,0, `xwininfo -root | grep Width | awk '{ print ($2/2)}'`, `xwininfo -root | grep Height | awk '{ print $2 }'`" (I've added line returns to make an otherwise massive one-liner readable.) I've bound this to a hotkey and it works, unless the window is maximized. Any ideas on how to fix this up?

    Read the article

  • How do I resize my emacs22 window?

    - by Evan
    I tried searching for answers but can't find any. I'm running Ubuntu 10.04 LTS and have emacs22 installed: ii emacs22 22.2-0ubuntu9 The GNU Emacs editor (Emacs 22) ii emacs22-bin-common 22.2-0ubuntu9 The GNU Emacs editor's shared, architecture dependent file ii emacs22-common 22.2-0ubuntu9 The GNU Emacs editor's common infrastructure ii emacsen-common 1.4.19ubuntu1 Common facilities for all emacsen I find that I cannot resize the outer frame (X window) of the emacs session. When I move the mouse to the corner of the window, it doesn't change into the resize icon.... help!

    Read the article

  • LXDE keybind "Raise" window not working

    - by Campa
    I am using LXDE* over Ubuntu Oneiric on a DELL Studio-XPS 64-bit machine and I can't get the keybind for raising windows to work. In my ~/.config/openbox/lxde-rc.xml I put: <!-- Iconify window: it works. --> <keybind key="A-Down"> <action name="Iconify"/> </keybind> <!-- Raise window: does nothing. --> <keybind key="A-Up"> <action name="Focus"/> <action name="Raise"/> </keybind> then I openbox --reconfigure, but no rasing windows keybindings. I could use the NextWindow action, but... not really the same. References: http://openbox.org/wiki/Help:Bindings http://openbox.org/wiki/Help:Actions *Nested question: how to know the version of installed LXDE?

    Read the article

  • run a script on window switch (on alt-tabbing into/outof a particular app)

    - by hwjp
    I'd like to run a script whenever I switch into or out of a window. Specifically, I have a "toggle touchpad on/off" script, which i want to run whenever I switch into or out of gvim. so, is there any kind of hook for that? even something that can be run whenever alt-tab is detected, as long as it can tell what window you're going from/to? I'm using gnome-shell, if that makes any difference. [edit] Have started looking into gnome-shell-extensions, which seem to be written in javascript, so might not be too hard to hack one together myself...

    Read the article

  • Problem with usercontrol scaling when added at run-time in .Net, WinForms

    - by Jules
    First here's the steps to replicate, then I'll explain what the problem is: (1) Add a usercontrol to a form in the construtor, after the InitializeComponent call. (2) Run the form. (3) Click a button to increase the form font size. All the controls within the usercontrol scale perfectly but the usercontrol itself doesn't. It's width and height are increased by way too much. To correct the problem, the usercontrol must be added before the InitializeComponent call. If it wasn't possible for me to add the usercontrol before InitializeComponent, is there any way for me to correct the scaling?

    Read the article

  • Scaling larger Image problem.

    - by krishna
    Hi, I m developing flex application, in which I want to Draw Image from User local hard-drive to the canvas of size 640x360. User can choose Image of bigger resolution & is scaled to Canvas size. But if user selected images of larger resolution like 3000x2000, the scaling take lot time & freezes the application until scale done. Is there any method to scale image faster or kind of threading can be done? I am using matrix to scale Image as below: var mat:Matrix = new Matrix(); var scalex:Number = canvasScreen.width/content.width; var scaley:Number = canvasScreen.height/content.height; mat.scale(scalex,scaley); canvasScreen.graphics.clear(); canvasScreen.graphics.beginBitmapFill(content.bitmapData,mat);

    Read the article

  • How to scale an image (in data URI format) in JavaScript (real scaling, not using styling)

    - by 103067513055141045393
    We are capturing a visible tab in a Chrome browser (by using the extensions API chrome.tabs.captureVisibleTab) and receiving a snapshot in the data URI scheme (Base64 encoded string). Is there a JavaScript library that can be used to scale down an image to a certain size? Currently we are styling it via CSS, but have to pay performance penalties as pictures are mostly 100 times bigger than required. Additional concern is also the load on the localStorage we use to save our snapshots. Does anyone know of a way to process this data URI scheme formatted pictures and reduce their size by scaling them down? References: Data URI scheme on http://en.wikipedia.org/wiki/Data_URI_scheme Chrome Extensions API onhttp://code.google.com/chrome/extensions/tabs.html The "Recently Closed Tabs" Chrome Extension onhttp://code.google.com/p/recently-closed-tabs

    Read the article

  • Android image scaling to support multiple resolutions

    - by tyuo9980
    I've coded my game for 320x480 and I figure that the easiest way to support multiple resolutions is to scale the end image. What are your thoughts on this? Would it be cpu efficient to do it this way? I have all my images placed in the mdpi folder, I'll have it drawn unscaled on the screen onto a buffer, then scale it to fit the screen. all the user inputs will be scaled as well. I have these 2 questions: -How do you draw a bitmap without android automatically scaling it -How do you scale a bitmap?

    Read the article

  • Proper chart scaling in Reporting Services 2005

    - by lastas
    I'm developing a simple bar-chart in Reporting Services 2005 with a stored procedure as data-source. The values in this graph can be both positive and negative, and can span a very big range, and hence I cannot specify any non-dynamic scale that will work for all scenarios. The problem I'm facing is that the automatic scaling pretty much sucks. I get no line to show where the zero-point is, and the y-scale labels are from top to bottom: 8818 -191181 -391181 etc etc... So my question is, what is the best approach to make the scale more adapted to human reading? Is there any guide out there? Does reporting services 2008 handle this better? Also, moving away from Reporting Services is not really an option. I realize how to put values and expression in the max, min, and the gridline interval fields, although its more of a question what expressions I should put there.

    Read the article

  • Confusion about TCP packet analysis terms

    - by Berkay
    I'm analyzing our network and have some confusion about the terms: this is the 2-packet output from source to destination. from these i have to get some features as describe, pls make me clear... packets with at least a bytes of TCP data payload: it seems tcp.len0; The minimum segment size (confusion is headers are included or or not) The average segment size observed during the lifetime of the connection, the definition: is calculated as the value reported in the actual data bytes divided by the actual data pkts reported. Total bytes in IP packets, should be ip_len value. Total bytes in (Ethernet) The total number of bytes sent probably related to frame.len and frame.cap_len these two terms are describes as, also make me clear about these two terms. frame.cap_len: Frame length stored into the capture file frame.len: Frame length on the wire

    Read the article

  • How can private IPV4 addresses get past iptables NAT (tcp RST,FIN)

    - by gscott
    I've got a router performing simple NAT translation using iptables iptables -t nat -o -j MASQUERADE This works fine almost all of the time except for one particular case where some TCP RST and FIN packets are leaving the router un-NAT'd. In this scenario I setup 1 or 2 client computers streaming Flash video (eg www.nasa.gov/ntv) At the router I then tear down and re-establish the public interface (which is a modem) As expected the Flash streams stall out. After the connection is re-established and I try to refresh the Flash pages, I see some TCP RST and [FIN,ACK] packets leaving the public interface (I assume as Flash attempts to recover its stream). I don't know how these packets can leave the router non-NAT'd

    Read the article

< Previous Page | 2 3 4 5 6 7 8 9 10 11 12 13  | Next Page >