Situation:
Varnish 2.1.4 lenny (latest stable) is running succesfully on Debian/Apache with mod_php.
I've got multiple Drupal sites running on this server and some are Varnish enhanced (e.g. using Pressflow). These sites work as expected.
The other sites, who aren't using the reverse-proxy, load fine when you enter the address or click around but take a really weird long time (30-45 secs) to load when you refresh the page using F5 or Ctrl+F5.
I don't think it's a VCL configuration thing; if I do a return (pass) immeditaley for one of these domains, the problem persists. Could this be related to this issue?
Are the sites that arn't using varnish still going through it?
For debugging, I would have a look at the net tab with firebug, it should give you an indication of what is taking its' time. It could be that the itial page load is fine but the other resources are taking a long time to get through.
Actually, I found out varnish 2.1.4 has some kind of problem with Apache's Keep-Alive functionalitiy (which was On). Turning it Off solved both problems!
Related
The title pretty much says it all. When I'm running HTTP(S) Test Script Recorder, one particular page becomes unresponsive - when I click on something it just reloads. The recorder itself is working fine, it is recording every step. And the problem is not proxy related. I've successfully recorded other pages of the same website. When I'm not recording, there are no problems.
It's a .NET 3.5 project.
The page itself has a lot of forms, file uploads, etc, but as far as I know, it should not interfere with recording or even more - with browsing the page. When debugging the project, no breakpoints are hit, so I assume that something gets lost before reaching the server.
Browsers that I've tried: tried FF, Chrome, Edge, IE.
Tried recording the web locally and online. Same thing.
Played around (reinstalled and whatnot) with certificates, didn't help.
Has anyone encountered such a problem? What could be done to fix this? I'm more interested in finding the solution, than a way around (blazemeter, badboy). Any help would be very appreciated.
EDIT: I tried recording with blazemeter and it worked. But when looking at view results tree I noticed that the request path and parameters don't change, even when in the HTTP Request Sampler they are different. So there's no solution yet
This often happens to me and what I've found is that JMeter changes the root certificate in the bin folder every week. So usually the HTTP traffic is fine but certain HTTPS traffic won't work. So make sure that your browsers are seeing the latest JMeter certificate and not using an old one that doesn't exist anymore. On Windows, Chrome and IE use the certificates in Internet Options, while Firefox needs to have it added to it manually.
It turns out that in Test Script Recorder HTTP Sampler Settings choosing Type: Java was all that was needed. I suspect that the issue was related to file upload being involved.
It's been a while now that i'm using Wordpress; and I know it quite well.
But this time, I've got a problem I've never seen :
My website freezes as soon as I try to go to the backend.
The frontend works, but when I go to /wp-admin, it stops working and eventually fires a "ERR_CONNECTION_TIMED_OUT" error.
Then the whole website (and FTP) is then down for about 5 minutes, then it comes back.
I don't know what as been changed since last time it worked.
Here's what I did try so far :
override my Wordpress folder with a fresh install
disable plugins (renamed wp-content/plugins to wp-content/pluginsOLD)
disable current theme (renamed wp-content/themes/my-theme to wp-content/themes/my-themeOLD)
edited wp-config.php file to enable debugging with WP_DEBUG, WP_DEBUG_LOG and SAVEQUERIES set to true; no debug.log is generated.
I also considered a server issue and waited for about 24hours. But my other websites (which are hosted on the same place) just work fine.
Any ideas about this ? I'm getting mad :)
My host had blacklisted my IP, that's why I wasn't able to access it.
I guess each time I was trying to access the backend, they did freeze my website to avoid an attack or something.
Well, this may be interesting for others. If I had knew it, I would have change my IP... :/
I have installed mybb in /forum and wordpress in /blog
Frequently, my wordpress blog doest not load (feels like the web server doesn't respond)
and at the same time, mybb also does not load
note: this happens very frequently but is fixed automatically after sometime
I have tried using Chrome and opera
----Chrome sometimes just keeps loading and sometime give "no data recieved"
----Opera simply keeps loading, loading and loading
First i thought this happens only with me but when i asked my friend to open my site, it happened with him too
If the server doesn't get hit very often, you may find that your web server is effectively turning itself off due to inactivity.
So when you visit it after a "long" period of inactivity, it takes a while while it load the application, rebuilds the cache etc.. ie the same as restarting the web application.
You can be sure this is not the problem, if you experience the "not responding" behaviour while you are actively browsing the site.
If you are aggregating contents then wordpress is not your choice. However I found cache plugins like quickcache can solve your problem.
Our customer experienced multiple errors on a demonstration to a customer of theirs.
The site is asp.net 3.5 based and has been running pretty well lately. They said 2 hours later site was ok again... (needless to say they are not happy)
The traces in the health log look very weird, it appeared to be as if incomplete pages were returning to the server for processing..
A fair of of the errors were errors generated by scriptresource.axd with bits of the page source showing up as part of the url... very very odd. I'd never seen this before...
I was talking to their lead tester who then told me this only happened during a webex (cisco sharing product) demo and was fine after...
Is it possible the webex session could have impacted the site http stream between client and server...
This is a known IE8 issue. I bet you will see Trident/4.0 for all "broken" requests in the user agent field. (Google for "IE8 4K bug")
You can either run a different browser or appply the latest IE8 patch (this issue has already been fixed)
We have an asp.net site running, which has been working fine for some time, but recently I have been experiencing some issues with IE8.
On posting some pages - mainly on our development server, although on staging too - we get an occasional "Internet Explore cannot display the webpage" error along with the button asking to diagnose connection problems. IE only seems to wait 10 seconds before timing out. I know that the page itself may take longer to load the first time (on dev and staging). So press F5 and everything then works fine.
Is there anything that should be done in the aspx page to tell IE to wait a bit longer?
I thought I had read that the default timeout supposed to be 90 seconds or something for browsers.
A bit more info:
It mostly happens on a POSTing a signup page, but that is just because I test that page and it starts the IIS App, makes the first connection to SQL and pre-caches some information. That first time the page can take 10-15 seconds to come back. IE8 times out after 10 seconds as it has had nothing back.
This happens on a dev W7x64 machine with 8GB RAM, as well as on a staging server WIN2008.
Having googled around a bit, some people are seeing the same problem, but no conclusive pointers to the problem or a solution.
It isn't a connection problem; everything works fine in Firefox, Chrome and even IE7; I have tried with add-ons disabled and resetting IE settings, still happens.
Ideas welcome.
Try this out
<httpRuntime executionTimeout="15"/> under system.web in the web.config
A 10 second timeout might be a usability tool in disguise. 10 seconds is a pretty long time. Js capable browsers are > 99% now, why not push it off over Ajax and poll until the long work is done, then redirect to the next page? You could show the user more useful progress info in the meantime. If for some reason the next page itself is the source of the slow down and cannot be separated from its slow parts, you could finish by precaching the next page then redirecting.
Check your Application pool Advanced settings in IIS. It may be lower than is normal. Maybe the Ping period? Mine is 30secs
I think that the long term solution is not actually related to timeout configuration.
You're saying that only the first request takes a long time, and it takes > 10 seconds, so you should issue a warm-up request first after installing your application in a way that the first request load is never experienced by the end user.