I managed a ASP.NET intranet site that use Integrated Windows authentication. It is deployed in many different countries for the staff in those countries. This week I have deployed to three countries that have slow connections to the rest of the internet. That said, the local network here is quite quick.
The site is relatively quick when a user continually make requests (clicks) to the server. But if the user in inactive for between one or two minutes, the next request is very slow. After which the request are quick again until there is a minute pause between requests.
What could cause this?
Also, I was wondering if the Integrated Windows Domain logging could affect the initial load time of pages. Does it query the active directory server or whatever?
Related
I have a problem with my website and I think it's related to IIS recycling the app pool or shutting down the app after 20 mins of inactivity. As it is a low traffic site, when I first browse to the site the initial load can take up to 30 seconds, then after that it is very fast. If i then come back to the site say a few hours later I get the slow load time again. I'm sure it's something to do with the app pool shutting down after 20 mins of inactivity?
Another problem is that I am hosting via a hosting company so have no direct access to manipulate IIS at all. Does anyone know how I can keep my app alive so I don't suffer from the initial slow load speed issues?
You could setup a service to issue a request every few minutes to your site which can double as monitoring your site to ensure it's still running. You probably don't have access to the application pool configuration I assume.
Our infrastructure team has configured a a load balancing using Radware. Basically we have 3 web server that are load balanced.
Before we go live I would like to test and make sure that load balancing is working. How do I test the following:
3 servers are load balanced and requested are evenly distributed. (Any automated tool exists?)
Asp.net InProc session are working.
You can test by first generating an artificial load on your site (with any one of a number of load generators). Then have a look at the Windows Performance Counters for each site: things like HTTP requests per second and CPU use would be reasonable high-level metrics.
Yes, there are automated tools, but they usually require quite a bit of setup, and the better ones charge a fee. Perf counters are fast, easy and free.
As #swannee said, InProc sessions won't work in a load-balanced scenario unless your load balancer is configured to use sticky sessions. It's better to use SQL Server sessions with load balancing.
FWIW, you can test your software in a "mini" load balanced scenario on a single server by enabling IIS web gardens (multiple worker processes), from the AppPool config dialog.
Can you look in the IIS server logs to see how many hits each server is getting?
http://msdn.microsoft.com/en-us/library/ms953324.aspx
Also, unless you are using sticky sessions, you are going to have problems with InProc session. It won't work on a server farm (unless as stated you have sticky sessions turned on). If you don't have sticky sessions, you'll be able to tell real quickly that your session is being lost between requests with just some manual testing.
Our organization makes a series of ping and advanced status pages. These pages are monitored by our load balancers so it can take out unhealthy nodes in the event one node loses a connection to a database server or the node itself is having issues.
Our ping pages spit out the server name that you're connecting to and the status. They are avaliable by the common server names themself, like server01.application.com/ping and server02.application.com/ping but more important, they all answer on application.com/ping.
Refreshing the page will show us a new connection (you can see the server name change).
To test load you could use WCat, it's not the easiest tool to setup and script but it works.
To test sessions. you'll need to build out some pages that you can do load testing on to test sessions
I'm building a web app using ASP.NET MVC and MS SQL Server CE and deploying it to a shared hosting. Every time I approaching my site after a pause (half an hour) it performs very slow. It takes about 40 seconds to load the first page. After this every page opens quickly.
Please advice what may cause the problem and how to solve it.
Thanks
Update: It appeared that after a pause in visiting (10-20 min) the site is loaded from the server hard drive not memory. That is why it was so slow. Possible solutions are increase site visits, move to a virtual server or ping the home page regulary.
My hosting provider has an option for pinging besides some solutions are described here c# ping a website? (keep-alive service)
If the worker process is not processing requests it will shut down the worker process after the specified number of minutes. (if configured)
Not sure what version of IIS you are using but look into "Idle Time-out (minutes)". IIS 7 specific info below...
http://technet.microsoft.com/en-us/library/cc771956(WS.10).aspx
As a paliative, set your app to recycle only after a long period of inactivity.
See Configure an Application Pool to Recycle at a Scheduled Time (IIS 7)
Hosted customers in IIS7 can use asp.net and System.Diagnostics to list all the system's process ID. They can also kill the ones that belong to their own application pools. Seems like a big security problems in IIS7 for shared hosting environment. Any suggestions on how to prevent normal users from accessing System.Diagnostics? How to limit it to administrators only?
Unlike with Windows 2003 and IIS6, many shared Windows 2008/IIS7 hosting environments provide their customers with dedicated application pools and Full Trust.
Whilst customers may be able to launch and kill their own processes (including their own worker processes), provided that the identity of the account that the site runs under is locked down then no real harm can be done. Also what would be the benefit to a customer having code that constantly kills their own application pool (other than to force a restart of a worker process to allow Application_Start type events to fire if you need to reload some settings there)?
I work for a shared hoster, we actually provide customers with the ability to start, stop and recycle their dedicated pools via our admin system, all they would be doing in code is pretty much the same thing.
The worst that can happen is that a customer launches a process that consumes large amounts of memory or an excessive amount of CPU (but then even their own ASP.NET code can run away out of control doing the same). We monitor our servers continuously for such anomalous behaviour and can track down the culprit within 2-3 minutes of being alerted. The end user will get a friendly warning and told not to do this again, if they do then their site is instantly shut down.
The only time I'd be worried if the hoster was running shared pools at full trust, but if they're doing that then they have a whole other security headache to overcome, process killing would be the least of their worries.
While tracing the active connection on my db i found that some times the connections exceeds 100, is that normal?
and after few minutes it return back to 20 or 25 active connection
more details about my problem
Traffic on the site is around 200 visitor per day.
Why i am asking? because the default MaxPool in the asp.net connection string is 100
Also i am using Connection in the website IIS
That really depends on your site and your traffic. I've seen a site peek out at over 350 active connections to SQL during its peak time. That was for roughly 7,000 concurent web users, on two web servers, plus various backend processes.
Edit
Some additional information that we need to give you a better answer:
How many Web Processes hit your sql
server? For example are you using web
gardens? Do you have multiple servers
how many if you do? This is important because then you can calculate how many connections you can have by figuring out how many worker threads per process you have configured. Assume worse case, each thread is running which would add a connection to the pool.
Are you using connection pooling? If so your going to see the connections stick around after the user's request ends. By default its enabled.
How many concurent users do you have?
But, I think your going after this wrong, your having an issue with no free connections available in your pool. The first thing I'd look for is any leaked connections (connections being held open for longer then they should). For example passing a data reader up to the Web Page, could be a sign of this.
Next thing is to evaluate the default settings. Maybee you should run a web garden which should give you more connections, or increase the number of connections available.
The last thing I would do is try to opitmize queries like in your last question. Let's say you cut those queries in half, all you've done is bought yourself more time until more users come onto the system, and your right back here, only this time you might not be able to optimize that query yet again.
You're leaving out some details making it difficult to answer correctly but...
It depends, really. If you're not using connection pooling then each time a page is hit that requires access to the database a new connection is going to be opened. So sure, it could be perfectly normal.
I would also look into caching. Cache pages, cache query results, etc. You might be surprised how many times you go back to the database to get a list of US States...