Search Results

Search found 17852 results on 715 pages for 'load balancer'.

Page 51/715 | < Previous Page | 47 48 49 50 51 52 53 54 55 56 57 58  | Next Page >

  • How to make Nginx fire 504 immediately is server is not available?

    - by Georgiy Ivankin
    I have Nginx set up as a load balancer with cookie-based stickiness. The logic is: If the cookie is NOT there, use round-robbing to choose a server from cluster. If the cookie is there, go to the server that is associated with the cookie value. Server is then responsible for setting the cookie. What I want to add is this: If the cookie is there, but server is down, fallback to round-robbing step to choose next available server. So actually I have load balancing and want to add failover support on top of it. I have managed to do that with the help of error_page directive, but it doesn't work as I expected it to. The problem: 504 (and the fallback associated with it) fires only after 30s timeout even if the server is not physically available. So what I want Nginx to do is fire a 504 (or any other error, doesn't matter) immediately (I suppose this means: when TCP connection fails). This is the behavior we can see in browsers: if we go directly to server when it is down, browser immediately tells us that it can't connect. Moreover, Nginx seems to be doing this for 502 error: if I intentionally misconfigure my servers, Nginx fires 502 immediately. Configuration (stripped down to basics): http { upstream my_cluster { server 192.168.73.210:1337; server 192.168.73.210:1338; } map $cookie_myCookie $http_sticky_backend { default 0; value1 192.168.73.210:1337; value2 192.168.73.210:1338; } server { listen 8080; location @fallback { proxy_pass http://my_cluster; } location / { error_page 504 = @fallback; # Create a map of choices # see https://gist.github.com/jrom/1760790 set $test HTTP; if ($http_sticky_backend) { set $test "${test}-STICKY"; } if ($test = HTTP-STICKY) { proxy_pass http://$http_sticky_backend$uri?$args; break; } if ($test = HTTP) { proxy_pass http://my_cluster; break; } return 500 "Misconfiguration"; } } } Disclaimer: I am pretty far from systems administration of any kind, so there may be some basics that I miss here. EDIT: I'm interested in solution with standard free version of Nginx, not Nginx Plus. Thanks.

    Read the article

  • How do I mirror a MySQL database?

    - by user45745
    I'm running two load balanced servers for one website, and I'd like the databases to be synchronized. Queries may be run on either of the two servers because they are both production sites, so the replication can't just work one way. It doesn't have to be in real-time, just fairly accurate so people don't notice a difference when they get switched to a different server.

    Read the article

  • How to load models in the extended MY_Router class in codeigniter

    - by askkirati
    I am not able to load models to the extended My_Router class in codeigniter. Below is my code: class MY_Router extends CI_Router { function MY_Router() { parent::CI_Router(); } function _validate_request($segments) { // Does the requested controller exist in the root folder? if (file_exists(APPPATH.'controllers/'.$segments[0].EXT)) { return $segments; } // Is the controller in a sub-folder? if (is_dir(APPPATH.'controllers/'.$segments[0])) { // Set the directory and remove it from the segment array $this->set_directory($segments[0]); $segments = array_slice($segments, 1); if (count($segments) > 0) { // Does the requested controller exist in the sub-folder? if ( ! file_exists(APPPATH.'controllers/'.$this->fetch_directory().$segments[0].EXT)) { show_404($this->fetch_directory().$segments[0]); } } else { $this->set_class($this->default_controller); $this->set_method('index'); // Does the default controller exist in the sub-folder? if ( ! file_exists(APPPATH.'controllers/'.$this->fetch_directory().$this->default_controller.EXT)) { $this->directory = ''; return array(); } } return $segments; } // Let's check if there are category segments $category_routes = $this->category_routing($segments); if($category_routes !== FALSE) { return $category_routes; } $user_routes = $this->user_routing($segments); if($user_routes != FALSE) { return $user_routes; } show_404($segments[0]); } function category_routing($segments) { $this->load->model('category_model'); if($this->category_model->category_exist($segments[0])) { //if only category if(count($segments)==1) { return array('category', 'category_browse', $segments[0]); } //category pagination if(count($segments)==2 and is_numeric($segments[1])) { return array('category','category_browse', $segments[0], $segments[1]); } //category upcoming if(count($segments)==2 and $segments[1] == 'upcoming') { return array('category','upcoming', $segments[0]); } //category upcoming pagination if(count($segments)==3 and $segments[1] == 'upcoming' and is_numeric($segments[3])) { return array('category','upcoming', $segments[0], $segments[3]); } //category top if(count($segments)==3 and $segments[1] == 'top') { return array('category','top', $segments[0], $segments[2]); } //category top pagination if(count($segments)==4 and $segments[1] == 'top' and is_numeric($segments[3])) { return array('category','top', $segments[0], $segments[3]); } } return FALSE; } function user_routing($segments) { $this->load->model('dx_auth/users', 'user_model'); if($this->user_model->check_username($segments[0])) { //only profile if(count($segments)==1) { return array('user','profile',$segments[0]); } //all friends if(count($segments)==2 and $segment[1]=='allfriends') { return array('user','allfriends',$segments[0]); } //all subscribers if(count($segments)==2 and $segment[1]=='allsubscribers') { return array('user','allsubscribers',$segments[0]); } //all subscription if(count($segments)==2 and $segment[1]=='allsubscriptions') { return array('user','allsubscriptions',$segments[0]); } } return FALSE; } } I have tried loading the models by using get_instance function provided by codeigniter but seems like it doesnot work. All i need is load the models in extended system library.

    Read the article

  • loading Data in VBA from a text file

    - by omegayen
    I am not very familiar with VBA but need to use it for a new software program I am using (not Microsoft related) I have a text file that has columns of data I would like to read into VBA. Specifically the text file has 4 entries per row. Thus I would like to load in the column vectors (N by 1). The text file is separated by a space between each entry. So for example I want to load in column one and save it as array A, then column two and save as array B, then column three and save as array C, and then column four and save as array D. This code snippet found below from http://www.tek-tips.com/faqs.cfm?fid=482 is something I found that can load in text to an array, but I need to adapt it to be able to save the columns as different arrays as specified above... Open "MyFile.txt" For Input As #1 ReDim Txt$(0) Do While Not EOF(1) ReDim Preserve Txt$(UBound(Txt$) + 1) Input #1, Txt$(UBound(Txt$)) Loop Close #1

    Read the article

  • How can i connect two or more machines via tcp cable to form a network grid?

    - by Gath
    How can i connect two or more machines to form a network grid and how can i distribute work load to the two machines? What operating systems do i need to run on the machines, and what application should i use to manage the load balancing? NB: I read somewhere that google uses cheap machines to perform this fete, how do they connect two network cards( 'Teaming' ) and distribute load across the machines? Good practical examples would serve me good, with actual code samples. Pointers to some good site i might read this stuff will be highly appreciated.

    Read the article

  • LVS TCP connection timeouts - lingering connections

    - by Jon Topper
    I'm using keepalived to load-balance connections between a number of TCP servers. I don't expect it matters, but the service in this case is rabbitmq. I'm using NAT type balancing with weighted round-robin. A client connects to the server thus: [client]-----------[lvs]------------[real server] a b If a client connects to the LVS and remains idle, sending nothing on the socket, this eventually times out, according to timeouts set using ipvsadm --set. At this point, the connection marked 'a' above correctly disappears from the output of netstat -anp on the client, and from the output of ipvsadm -L -n -c on the lvs box. Connection 'b', however, remains ESTABLISHED according to netstat -anp on the real server box. Why is this? Can I force lvs to properly reset the connection to the real server?

    Read the article

  • Fortigate - Accessing a Virtual Server address from several interfaces

    - by Jeremy G
    I am setting up a new application in its own DMZ on our Fortigate 300C firewalls. I have defined a load-balancing configuration for part of the application, and this works fine for traffic coming in from our internal network. However, I would also like this application to be reachable from other DMZs, for inter-application traffic, and from the SSL VPN interface. I can't seem to define the required policy, and it seems this is due to Virtual Servers being bound to the client interface on the Fortigate rather than the server interface (and so my virtual IP is not accessible from any of these other interfaces) Does anyone have an idea how I might go about this ? I guess I could create other virtual IPs for each interface, but this gets complicated to handle as clients need to change the address they use depending on how they are connecting. Thanks, Jeremy G

    Read the article

  • scalability with nginx, passenger, ruby on rails setup

    - by Dani Cela
    Hey guys I had a question regarding scalability for my RoR application. We have been optimizing our application over the last few days and after running blitz.io, notice that our application times out after maybe 1000 hits in 30 seconds we experienced massive timeouts. In the 1 minute test apparently 74% of users would have timed out. Look at the performance of my website: http://blitz.io/report/1c8eb2f395a5eadeabd62fd831ada9e5 Not saying that our website will in any way experience this now, but I wish to design the infrastructure to handle this. What is normally done in this situation? Currently we have one web server and one database server. Would load balancing be the route to go?

    Read the article

  • loadbalancing with difference nginx location context and backend server context

    - by robinmag
    Hi, I used nginx and upstream module for load balancing with the following config upstream lb { server 127.0.0.1:8080; server 127.0.0.1:8081; } server { listen 88; server_name localhost; location /cas/ { proxy_pass http://lb; proxy_redirect off; proxy_connect_timeout 2; proxy_set_header Host $host; proxy_set_header X-Real-IP $remote_addr; proxy_set_header X-Forwarded-For $proxy_add_x_forwarded_for; } } the problem is the "location /context/" have to match to the context of backend server so when i request localhost/context/index.html then nginx routes it to 127.0.0.1:8080/context/index.html or 127.0.0.1:8080/context/index.html. Is it possible to have difference backend context and nginx location for example with "location /" nginx will routes the request to 127.0.0.1:8080/context/index.html or 127.0.0.1:8080/context/index.html Thank you.

    Read the article

  • loadbalancing with difference nginx location context and backend context

    - by robinmag
    Hi, I used nginx and upstream module for load balancing with the following config upstream lb { server 127.0.0.1:8080; server 127.0.0.1:8081; } server { listen 88; server_name localhost; location /cas/ { proxy_pass http://lb; proxy_redirect off; proxy_connect_timeout 2; proxy_set_header Host $host; proxy_set_header X-Real-IP $remote_addr; proxy_set_header X-Forwarded-For $proxy_add_x_forwarded_for; } } the problem is the "location /context/" have to match to the context of backend server so when i request localhost/context/index.html then nginx routes it to 127.0.0.1:8080/context/index.html or 127.0.0.1:8080/context/index.html. Is it possible to have difference backend context and nginx location for example with "location /" nginx will routes the request to 127.0.0.1:8080/context/index.html or 127.0.0.1:8080/context/index.html Thank you.

    Read the article

  • Looking for the best ec2 setup for 3 sites totaling in 1.5 mil in traffic monthly

    - by john h.
    I am looking to consolidate our current aws setup of 2 Large ubuntu ec2 servers and 2 large RDS server for our 3 websites that have a total of about 1.5 million hits a month and increasing every month with the majority of traffic (1 mil) to one forum site in the group and the rest of traffic to an ecommerce site and a small wordpress site. So here is my question/thought? Would it be better for us to combine the two ec2 large servers to just one and same with the 2 RDS servers so we run all three sites off one large ec2 and one RDS. -or- Should we setup maybe 2-3 smaller ec2 servers load balenced and a single RDS. -or- Something completely different setup? One concern is that if one site crashes it takes with it the others. It happened in the past but I am pretty sure its because of the forum software and not the server setup. -john

    Read the article

  • Good Books About Scaling Up Databases/Servers/etc.?

    - by Mehrdad
    I've applied for an internship at a startup company that expects its user base to grow by a large factor in a small amount of time, and so part of their project is to scale everything up so that they're ready: handling more/larger requests efficiently, handling server failures, load balancing, getting more JavaScript to run faster on the client computers, etc. Part of my job will also be figuring out what to do, so it's not obvious what my exact task will be at the moment. I was told that I should start reading up a little more about this so that I would have a little bit of an idea of what to do. What are some good books for me to read on this topic? I have a little bit of experience with the usage of MySQL (and also a little experience with web development), but in no way do I claim any knowledge on the internal workings of databases or distributed systems, so I might need readings more on the introductory side.

    Read the article

  • Loadbalance UDP traffic with session affinity and way to take servers in & out of rotation

    - by William
    What is the best way to go about load balancing UDP traffic among a whole bunch of servers, while keeping session affinity based on the users' IP? I need to also be able to take servers in and out of rotation for new clients, so when they join for the first time, they get put on a server in a list of available servers, and clients already connected would stay connected to their specific server. I have written the software to maintain a list, but I can't seem to find anything that would perform this functionality. If you need the context, this is to facilitate game tournaments for Minecraft: Pocket Edition, which is done with UDP traffic, I cannot change the protocol. And, because tournaments open and close, I need to be able to place players on their proper servers. Performance is also a priority, I have a program to do this but it is very bloated and slow. Thanks for any help! William

    Read the article

  • LVS TCP connection timeouts - lingering connections

    - by Jon Topper
    I'm using keepalived to load-balance connections between a number of TCP servers. I don't expect it matters, but the service in this case is rabbitmq. I'm using NAT type balancing with weighted round-robin. A client connects to the server thus: [client]-----------[lvs]------------[real server] a b If a client connects to the LVS and remains idle, sending nothing on the socket, this eventually times out, according to timeouts set using ipvsadm --set. At this point, the connection marked 'a' above correctly disappears from the output of netstat -anp on the client, and from the output of ipvsadm -L -n -c on the lvs box. Connection 'b', however, remains ESTABLISHED according to netstat -anp on the real server box. Why is this? Can I force lvs to properly reset the connection to the real server?

    Read the article

  • Round-Robin DNS in mobile networks

    - by k7k0
    After reading load distribution alternatives and giving my limited skills on the area I'm biased toward round-robin DNS strategy. From what I understood, one key aspect of DNS Round-Robin is setting a low TTL value, avoiding caching. My main concern is that all my traffic comes from mobile networks, almost 30% of that comes from t-mobile 3G. Some questions: 1) Is there a chance that almost all clients on the same mobile network will be redirected to the same IP in the TTL frame? That would kill the distribution technique. 2) If I choose a really low TTL (zero or one). That impacts directly over client performance? It does a DNS miss every time or it's a setting that only impacts on DNS servers? Any help would be much appreciated. Thanks

    Read the article

  • How to limit disk performance?

    - by DrakeES
    I am load-testing a web application and studying the impact of some config tweaks (related to disk i/o) on the overall app performance, i.e. the amount of users that can be handled simultaneously. But the problem is that I hit 100% CPU before I can see any effect of the disk-related config settings. I am therefore wondering if there is a way I could deliberately limit the disk performance so that it becomes the bottleneck and the tweaks I am trying to play with actually start impacting performance. Should I just make the hard disk busy with something else? What would serve the best for this purpose? More details (probably irrelevant, but anyway): PHP/Magento/Apache, studying the impact of apc.stat. Setting it to 0 makes APC not checking PHP scripts for modification which should increase performance where disk is the bottleneck. Using JMeter for benchmarking.

    Read the article

  • How to optimize mysql databse for serving 1000s of requests at a time ?

    - by Bilal
    How to optimize mysql databse for serving 1000s of requests at a time ? for a site like: linksnappy.com Is it possible to configure 2 seperate mysql servers into one load balancing server? like if one of them is overloaded switch to the next one ? same question for the http requests handling server. Another question : what kind of server do I need to server 1000s of requests at a time ? (http server) you can see the kind of site Im talking about is a download site. the server just dies when we have too many download requests. we currently have intel xeon quad core 2.4ghz with 4gb of ram.

    Read the article

  • Drbd Primary/Primary + iSCSI: accessing to different files avoids split brain?

    - by Eddie C.
    I have a question / curiosity about split-brain on a Drbd Primary/Primary configuration. Supposing two nodes (hosts), host1 and host2 configured with Drbd Primary/Primary and two different shares (NFS, CIFS o iSCSI) of a replicated area (saying /drbd) /drbd/file1.data /drbd/file2.data If a pool of client would access only by host1 share reading and wrinting only file1.data and another pool only by host2 share to file2.data, this scenario should avoid split brain situation in case of one node failure or it's just a conjecture? The final purpose is load balance between the two nodes in normal condition and collapsing to one node only in case of failure. Thank you! Eddie

    Read the article

  • How can I making Twitter, Facebook and Reddit share buttons load last?

    - by Daniel Bingham
    I have a website with a number of pages that sport twitter, facebook and reddit share buttons. They take forever to load and until they do the rest of the page doesn't load. So how I can make them load last? Currently, they are loaded something like this: <div class="item"><a href="http://twitter.com/share" class="twitter-share-button" data-count="vertical" data-via="FridgeToFood" data-related="danielBingham:Recipe and update tweets from Fridge to Food.">Tweet</a><script type="text/javascript" src="http://platform.twitter.com/widgets.js"></script></div> <div class="item"><script src="http://connect.facebook.net/en_US/all.js#xfbml=1"></script><fb:like layout="box_count" width="40"></fb:like></div> <div class="item"> <script type="text/javascript">reddit_target='recipes';</script> <script type="text/javascript" src="http://reddit.com/static/button/button2.js"></script> </div> They are in a div called "shareWrapper" and are loading to one side of the page. The buttons load where ever the script code is placed. As far as I know, I can't place the script code at the bottom of the page and move the resulting buttons after the fact. I want them to appear near the top, which right now means they are stopping everything below them from loading for several seconds. I tried loading them using javascript, but using JQuery's $(document).ready(), but that failed. It seems to leave the page in some sort of loading loop from which it never emerges. Are there other ways to get these to load last?

    Read the article

  • SSL Certificates, two-way authentication and loadbalancers

    - by 5arx
    We're looking to implement two-way authentication with client certificates for a privileged subset of our application users. The idea will be that if a certificate is detected the user will be asked for an additional password/PIN and that will be used to verify the certificate and user. Ordinary users will continue to authenticate themselves via the standard login mechanism. Our production environment (hosted by a well-known company) comprises load-balanced application servers and I'm unclear as to how this set-up will handle the certificates and I'm not certain if there are any pitfalls I should be aware of. I would very appreciate some thoughts, comments or real-world advice on the subject.

    Read the article

  • 250k connections for comet with node.js

    - by Nenad
    How to implement node.js to be able to handle 250k connections as comet server (client side we use socket.io)? Would the use of nginx as proxy/loadbalancer be the right solution? Or will HA-Proxy be the better way? Has anyone real world experience with 100k+ connections and can share his setup? Would a setup like this be the right one (Quad core CPU per server - start 4 Instances of node.js per Server?): nginx (as proxy / load balancing server) / | \ / | \ / | \ / | \ node server #1 node server #2 node server #3 4 instances 4 instances 4 instances

    Read the article

  • How can I determine Breaking point of my Web application using JMeter?

    - by Gopu Alakrishna
    How can I determine Breaking point of my Web application using JMeter? I have executed the JMeter Testplan with different concurrent users load. EX. 300 users(0% error), 400 users(7% error in a sample, 5% error in another sample), 500 users(more than 10% error in 4 out of 6 samples). At What value of % Error, I can say system reached the Breaking point.I used concurrent users 300, 400, 500 in a PHP website. Should I consider any other parameter to determine breaking point. How many maximum concurrent users my application can support?

    Read the article

  • Apache stops serving requests when connections increase

    - by Gunjan
    The values for MaxClients, ServerLimit etc parameters are quite high (4000). Available RAM on the server is high too (~8G). Load average remains below 1 on a 24 core CPU. But when the number of visitors on the website increase apache just stops serving requests. The apache error log is blank and access log shows no more requests coming in. Restarting apache makes it work again until the number of requests increases again. Any ideas where to start looking? UPDATE Getting the below errors in apache error log on running it with LogLevel Debug [info] server seems busy, (you may need to increase StartServers, or Min/MaxSpareServers), spawning 32 children, there are 479 idle, and 1027 total children

    Read the article

< Previous Page | 47 48 49 50 51 52 53 54 55 56 57 58  | Next Page >