pulling my hair out with this one, not sure what the problem is.
I have two desktop machines, both running XP SP2 and one has IE7 the other IE6.
Other than urlblast, nothing else is installed. Both machines are connected directly to an ADSL modem/router.
Recently when parsing the data, i saw a sudden jump in page load time for the IE6 machine. Nearly all of the un-primed cache tests failed and out of the 400 (or so) primed cache tests only 13 were successful.
Now when i let the tests run and observed the tests running, IE6 opens and the page starts to load. However on the first load, there is a JS error, but the browser says “Done” in the status bar. urlblast however times-out the browser, takes a screen shot and fails the test. The IE6 machine is reporting an average of 4 URLs or even N/A a day.
So it was working, and now its not.
If i log on as user1 and open the browser i can navigate the site just fine.
The IE7 machine is working just fine.
FYI
I’ve killed and recreated the “user1” user and made sure its a machine administrator. urlblast runs from the administrator account and the machine is logged in.
Is it just that site you see problems with or are you having problems across the board? google.com is usually a pretty good baseline to test against.
For the IE6 tests do you get a document complete at all (on the failed tests)?
Does the browser hang (no results at all) or does the test jjust timeout with a 9999x error code?
Have you been using 217 the whole time?
We’ve seen problems before with 3rd party javascript where they prevented the onLoad from firing and caused similar timeout issues and since it wasn’t something we changed it was pretty hard to track down. I have access to an IE6 test machine at the office - I’ll give the page a shot from there and let you know how it goes.
Looks like the browser is getting locked up (not a simple case of no document complete). Debug builds of pagetest can spit out a bunch of information. I’ll take a lookk and see if I can track it down.
i logged into the machine as user1 and started up IE6 and browsed to the url. all seemed fine. But when i closed the browser i got an error from IE6 (it had crashed).
Thinking that IE6 may have been having trouble with that users profile i logged out as that user and back in as administrator. I deleted the user user1, and chaned the urlblast ini file to create a user of the form “urlblastn” and restarted urlblast. the user urlblast1 was sucessfully created, IE6 was started and navigated to google and closed, then reopened and navigated to google (awesome i thought, its fixed), but then the browser window opened again to visit the autotrader site, nothing was drwan on screen, and the browser closed. The browser started again, and again nothing was rendered (i could see the call to the url in the status bar of IE6).
Then without warning the machine rebooted.
I have urlblast set as a startup short-cut so once windws was back up, it did exactly the same thing.
I grabbed the logs (attached), what does “Too Many Sequential Errors” mean?
Cheers,
Stuart[hr]
Umm ok, i’ve just fixed it, and its bizzare.
The machine is one of our old “stock” desktop PCs that was ready to join the great scrap heap, ad the IT guys wiped it clean and put a new corporate XP image on there (it included SP2).
Now after reading around on the internet about IE6 crashes i found a few articles about a particular culmative secuirty patch for IE6, so i went to the MS update site, to find there was only one update, windows genuine advantage. I installed that, and then lo and behold it all started working!
And of course as soon as i installed that MS updates reports there is 127 critical security updates required for IE6!! I haven’t installed any of them, as one of the criticisms i have received is that the site crashes IE6.
Wow, thanks for tracking that down. That’s going to be a “good to know” for any IE6 testing.
The “Too many sequential errors” is some protection that urlblast has/had in place to reboot the PC if it looks like it is having networking or other problems. After 3 hard failures (not 404’s but things like a DNS lookup failure for the base page) it would reboot. This was put in placce a few years ago because we had some test machines that would sometimes come up and the network would not negotiate but a bounce would fix iit. I removed the code a couple of weeks ago because it can do more harm than good but haven’t cut a new urlblast release yet.
