I have an application pushing data on long WAN links. The application runs on RHEL4.
It is not able to utilise the complete available bandwidth. I was told that kernel needs to be tuned for "long fat" links, so that the system starts utilising complete available bandwidth. I found a link [URL] however it is on FreeBSD.
get me understand the short range and the long range links from routing (and routing protocols') point of view.
View 6 Replies View RelatedRalink's drivers for wireless adapters [url] with .dat-file to configure default properties of used WiFi adapter (at my case - RT2870.dat).
Example is below.
I'd want to fine tune adapters (I have two of them) to get maximum performance for streams between two WiFi clients connected to my router (JWNR2000 at my case).
At the moment I have changed WirelessMode=6 to force 802.11n mode (and it does work). Speed between WiFi-clients is about 3MB/s now. I want more How to?
Code:
I have an enormous quadcore machine with 16gb ram and dual gigabit NICs. It used to be for MySQL but we have upgraded the whole database infrastructure so now this server is left floating. I had the great idea of turning this into a reverse-proxy (using apache mod_proxy) and it really handles a ton of requests. But I have a feeling that we are not getting the most use out of what it can offer.
Our traffic consists of a few thousand very small (less than 10 byte) ajax calls per second, and frequently I find we are running out of kernel allocated network stack to handle all the requests. Often we get the kern.log warning "possible SYN flooding on port 80. Sending cookies." and other things like this. Obviously we are not getting SYN flooded, we just have very high demand.
So far I have found a few kernel tuning guides to tell the kernel to allocate more of the base system memory for networking but every guide I have found has been for the purpose of increasing the performance between WAN links (direct backbones between offices etc) and usually with very large file sizes being the priority. One such example (and great) write up is here:
cyberciti.biz/faq/linux-tcp-tuning/
I was hoping some people could provide further input, such as along the lines of disabling nf_conntrack (to speed up socket set up/tear down time) or anything that will speed up a high throughput proxy like mine. Any links to studies or benchmarks between different configurations or hardware gets extra points!
At my Uni, we use a web-based login for our internet connections. Its based off of Cisco, and every Wednesday night every computer on campus must re-enter their credentials to use the network.
Normally on my several computers I simply pull up the Terminal, point links to google.com using
Code:
And enter my credentials when Cisco redirects to the login page.
Literally, the process is
Code:
Then ENTER to accept the redirect, down arrow to skip over the logo image, USERNAME, ENTER, PASSWORD, ENTER, ENTER.
Naturally, this is EXTREMELY time consuming, as I have about 5 computers located around campus and must physically walk to the machines and login every single week.
My question is, How would I formulate a program that does the following;
1) checks for connectivity (i.e. is able to reach/resolve to the greater part of the internet) and
2) automatically fills in the credentials on the links login page?
I have a personal wiki of notes, with now thousands of links in markdown format:
[link text](http://example.com)
but now that fckeditor is available for mediawiki (very beta), it has become much better to just stick with wikitext format. There are only a few conversions to do: tables, links, and bulleted lists. The lists are a fairly simple regex and fckeditor magically reformats the tables, so all I'm left with is the links. But I'm not a regex master. How do I reformat code...
just start Ubuntu 9.04 said: File system chek failed a long is beging saved /var/long/fsck/checkfs if that location is writable Please repair the file systmen manually A maintenance shell will now be started Ctr+ D terminate this shell and resume system boot. Give root password for maintenance or type Control +D to continue. I did Ctr+D , and after login said , that can not find /home. I starte with the live cd:
[Code]....
My hosting server is running Linux / Apache. It would be very nice to be able to link some files (preferably hard links, but symbolic links also would help), but haven't a clue how to do so. I would be willing to write a server side php script if that would do the trick.
View 4 Replies View RelatedI'm trying to setup a linux box with 3 NICs (2 WAN links and 1 LAN). All http traffic (port 80) should go to WAN 1 via squid proxy and the rest to WAN 2. I already setup MASQUERADING in iptables and I already configured port 80 to redirect to port 3128 for squid. My default gateway is WAN 2. But the problem is squid uses the default gateway - WAN2. setting up the iptables / routing for squid to use WAN 1?
View 7 Replies View RelatedI'm trying to get granular with TCP, one of the answers I can't seem to find is. When a client server connection is lost without closing, how many duplicate ack's are sent until the session is closed. I've kind of figured out the the sender will continue to retransmit a segment for about 9 mins if not Ack'd. going from 1.5,3,6,12,24,48 and the continuing with 64 seconds up to around 9 minutes before is resets the session. but whats happening at the other end.?
View 1 Replies View Relatedwe having more than 5000 users and will have 7 squid proxy servers with high end configuration upto 4gb ram n 320*5 HD in rhel4&5 most of them complaining that at peak hours their browsing speed is slow but we are having 1gbpgs link at peak hours i.e when established connections r flowing more than 550 browsing gets slow how to do fine tuning are squid is only responsible to access upto 600 connections
View 2 Replies View RelatedWe need an official IO tuning guide for ubuntu similar to the one given here [URL].
View 4 Replies View RelatedHow to speed up the menu "fade" (or get rid of it all together) in desktop effects? I need a composite manager for docky but the slower response on menus annoys me as they have to fade in and out. Anyone know which key in apps -> compiz I should change? Is it one of the plugins or one of the general keys?
View 5 Replies View RelatedEssentially, what I am attempting to do is set up an older PC with a Radeon 9600 video card chipset in it with xubuntu 9.10. I would like to attach this PC to my rear projection tv (Sony KP-48S75) via an S-video cable and use this TV as the primary and only monitor for the system. At one point, I had Ubuntu 8.04 hacked onto this same hardware fulfilling this role decently. However, due to some carelessness on my part, the system became unstable and I decided to start anew (random crashes, lots of X freezes, dependency issues, screwing too much with my sources.list and so on). So, that said, I installed Xubuntu 9.10 on the PC and had it hooked up to both an LCD monitor and the TV via S-video. Upon boot, both outputs worked, though X only rendered on the LCD and not the TV. Thus, I dug and researched and found that by putting a bunch of xrandr commands into my gdm start script I could finagle the TV into displaying my output, mostly.
The commands I used to get the S-video out work (and, therefore, pasted into my /etc/gdm/Init/Default script) follow:
Code:
xrandr --output S-video --set load_detection 1
xrandr --output S-video --set tv_standard ntsc
xrandr --output S-video --set tv_horizontal_position -2
xrandr --addmode S-video 800x600
xrandr --output S-video --mode 800x600
xrandr --output S-video --rate 60
xrandr --output VGA-0 --off
You'll notice I do a few things in those commands. First, I set load_detection to 1 in hopes of enabling automatic S-video connection detection (this isn't working). The next interesting bit is my commanding the horizontal position. The reason I had to do this is, because, upon getting the S-video output to render, the picture rendered to the right-hand side of the screen, deprecating the furthermost bits (for instance, I cannot see the shutdown icon in the upper right hand corner). Setting the horizontal position to -2 allowed me to shift the display to the left a bit, but, to my surprise, the right hand side still deprecates and I am just left with a black column on the right hand side.
Question 2 has to do with the screen indexing. If I run a xrandr command with no options, I see VGA-0 indexed as Screen 0. I also see the S-video listed as disabled (even though I see it rendering on the TV right in front of me). Finally, I notice that, under the VGA-0 section, there is a screen size listed, as well as several resolution modes with many parameters following them (like 800x600 72.2*+ 75.0 .... some other stuff). However, under the S-video section all I see is one mode with one small bit after it (800x600 72.2*). This smacks of suspicion to me and makes me think it might be related to the right side of my screen being deprecated. My wager is that I need to tweak some modelines in the S-video output so that it is more in sync.
However, I do not know the syntax for modeline editing and, since the S-video screen isn't even detected, I am not sure I really can edit these options. Does anyone know why my S-video output would continue to be listed as disabled even when its running? Furthermore, can anyone tell me why no screen index is given to the S-video output (I only see screen 0 attached to VGA-0, no screen 1 for S-video)? I think I need the screen indexed properly in order to use the xvattr command later in order to allow xv overlays to function on the proper screen. Finally, I do have a xorg.conf file loaded in /etc/X11 that is mostly hacked together stuff I used trying to get this to work in the first place. Would this conflict with the xrandr setup? Which options override which?
I was so sure it must be easy to set up the file permissions like I want, but obviously it isn't. Here is what I want to have: I have a samba fileserver (only Windows clients connecting) with a common share. Everybody is supposed to be able to put files on the share, but no one should be able to delete files which do not belong to oneself. I want several users to put files in a common directory, so they need write access to the directory, right?
But all the same they should not be able to delete files from others. So they are not allowed to have write access to the directory, right? It all works fine with modifying etc. but still I can delete files I am not supposed to do. I learned already that permission to delete is bound to the permissions of the containing folder and that sticky bits might help. But either they don't or I didn't get it right until now.
I have a Asus 54G43-Pro mobo with Intel GMA G43 video chip on it. It works fine in Ubuntu and using the same display with the same settings, under Maverick in comparison with Lucid, the video is a little more sharp and with more contrast. I have dual boot with Windows and in Windows it is even less sharp and contrast.
In Windows I can tune the sharpness, contrast, etc but I was not able to find a simlar features in Ubuntu. I have checked the Synaptic and even installed some Intel GMA files but I still have no access on tuning the video driver. Is this possible in Ubuntu and possible to be hidden somewhere or I can just forget about it?
I am currently reading [URL] its an old document from 2006, so i am not really sure how much information in it is still valid.
View 1 Replies View Relatedi need to get gedit to highlight lisp source code as well as provide auto indentation.(because as you may know , lisp programming without auto indentation is IMPOSSIBLE) now so far i've searched the web and tried some "solutions" but didn't work so far , for instance this post here promises syntax highlighting but suggesting to install 2 files , unfortunately it seems to do nothing.
View 12 Replies View Relatedwe are setuping high-performance web farm with 4 ubuntu servers (all 10.14 LTS 64bit, i5Core CPU, 16GB Ram). One server is "database" server with mysql installed, three servers are "webservers" with tomcat installed. All three webservers have public IP and are connected directly to internet by eth0 (onboard network card) and are connected to switch that is connected to database server by eth1 (pci network card). All servers has stardard instalation with no special modules/programes installed.
Web servers are already under load, but only on about 10 requests/minute. After several hours in production, strange thing happened - one of the web servers (each time another one, so it is happening to all of them but not at once) starts to have pings to database (that is connected directly on LAN) = 100ms (other two servers still has like 0,02ms). Pings stay high until we do /etc/init.d/networking restart, after that it is fixed and pins are again low again. Another strange think is, that when we let ping running for some time, it is slowly dropping from 100ms to 1 ms and then again jumps to 100ms (exactly to 100ms always) and again starts dropping - so it is cycling over and over again (one cycle takes about 2 minutes).
We were trying reconfigure network, play with switch (the same situation was happening even without switch when there was only one webserver and one database server directly connected), disable IPv6 but nothing helped. The only thing that we discovered is that it has to be caused by network configuration.Do you have any idea what can cause it? We went through systems logs but found nothing suspicious (at least according to our understanding).
I am having a problem with internet browsing. Whenever I type the address of any webiste like Google
or URL...or any other, the firefox gives message at "Status Bar" that "Looking up google.com" or "Looking up yahoo.com" etc for every website for about 20-25 seconds and than launches the website successfully. Why this is happening and how can I correct it?I also tried this in Opera and Google Chrome and same problem occurs in them.Moreover, if I ping any domain inside terminal, it instantly starts pinging (with ttl 51) and recieving data from that domain resulting that the problem is with browsers.
I have a RHEL 4 server delivering long messages to several RHEL 4 clients. They work fine.I am trying a RHEL 5 client running the same client application as the RHEL 4 clients. The RHEL 5 client looses most messages. have tried with two other clients running CentOS 5 with similar results. The server indicates write failure or blocked. The messages are typically 50 kilobytes. I have changed many parameters in sysctl.conf Has anyone encountered similar problems?
View 3 Replies View RelatedI was setting up an Ubuntu server for a client who runs a rather old version of Irix on an SGI O2 machine (he uses some design software that does not run on newer hardware). This version of Irix can only use NFS V2 or V3, so I setup the exports file on Ubuntu to use nfs3. However, when the file system was mounted on Irix, everything (including flat text files) was mounted as a directory, i.e. everything in the Ubuntu directory was mounted as a circular link to its' self After _much_ faffing about, I found that using nfs2 in the exports file fixed the problem, and the client was a happy camper again.
View 1 Replies View RelatedMy netowrk interface:
My iptable T1:
http traffic not follow the iptables. sometime via eth1 or eth2
After squid.conf added "tcp_outgoing_address wan1's gateway" it worked correctly
Why? what squid do?
The http traffic:
Why iptables can't not control the traffic
I want creat the tables T2 =>for load balancer
How to make squid route follow this table
I am using Ubuntu-10.4 and postgresql-8.4.Now I have to change my SHMMAX variable for performance tuning.I have 16 GB.change it permanently and what is the suitable value for the DATABASE Server.
View 1 Replies View RelatedI've been trying to set-up/tweek these param's in Apache to a setting suitable to the server for the amount of memory. When i look around some people say, hey.. just look at the memory used per process and then divide that by the amount of memory available and you get the number of processes that can be handled by Apache in one go before it starts swapping.
Well, for this i'd done this calculation and for me it turned out to be 200 approx concurrent connections. Well funny thing is, out MySQL server had a slow down so the Apache servers were running at approx 450 concurrent connections and weren't swapping in memory etc (still had 600MB free not including what was available in Buffers/Cache - 'free -m')... Thus if i had have set the limits to 200 then people would have been not getting to the site, so i'm kind of pleased that i happened to not have the time to set this yet.
My intention is to have EngineX on the frontend handling all static files, and Apache + mod_php in charge of handling PHP requests. I have optimized nginx to be as efficient as possible in handling static files, but have very little experience with Apache. Since Apache would only be processing PHP requests, would the standard Apache optimization guides suffice or would it best to configure it differently?
PS: this is a dedicated file server, the database is hosted separately.
I was on a car trip earlier, but we were stuck in traffic. The stop-go stop-go kind of BS. I pulled out my 9.10 laptop and figured I'd scan for wifi networks and see how good America was at securing their networks. (I work in IT support so it's one of those things I preach to no end yet people rarely listen).
I was using wifi radar, and picked up a truckload of networks every time we'd stop. But something didn't sit well with me. When I click on the network manager icon in the top gnome panel to see what networks were available, it never seemed to refresh.
Example - the first 5 networks I saw were 1, 2, 3, 4, 5. We would stop-go more and more and miles later when I'd see more networks in the area show up in wifi radar, yet hit my network manager, I'd STILL see networks 1, 2, 3, 4, 5 and not 30, 31, 32, 33, etc like I expected.
What kind of refresh time or cache or whatever does network manager have? I just didn't understand why I was seeing the original networks from 15 minutes/miles ago and it never seemed to update.
My problem is that every time I use the Ethernet connection (I don't have wireless card) there is a delay of around 30s, even though my download speed is not affected at all once the connection is "established" , no matter what application I use (Synaptic, ssh, firefox, google chrome,wget, update manager, etc.). A weird thing is that my ethernet connection always appears as active.
I'm sure I'm not the only one who has experienced this issue, since I've read the same problem on other forums, but no one has come up with a solution. I hope that there will be a solution at last by this time.
I've been around with this problem for about 1 month, I've tried all sorts of tweaks that involve disabling ipv6 and nothing happens. I don't really know what triggered off this issue because it all started all of a sudden and I don't remember having installed any important update by the time this happened (I did a clean installation of Ubuntu 10.04 32bits on the release date).
I have a Samba share that contains a symbolic link and when I try accessing it from the WinXP machine it denies permission. If I access it from the Linux account, it goes in with no problems. Is there a certain setting that needs to be set or enabled or is this just one of those things with Samba?
View 5 Replies View RelatedI've created two samba shares, /media/disk1 and /media/disk2 say the structure is like this:
/media/disk1/dir1
/media/disk1/dir2 links to /media/disk1/dir1
/media/disk2/dir1 links to /media/disk1/dir1
so I can access /media/disk1/dir1 and /media/disk1/dir2 from samba share, but I cannot access /media/disk2/dir1, on Windows 7 it throws:
Quote:
[Window Title]
Location is not available
[Content]
N:dir1 is not accessible.
Access is denied.
[OK]
In conclusion, symbolic inside the same drive is okay, but when cross-drive links happened, I cannot access them..