Version 3.0 Beta Test FeedBack

Dave,

Any superuser will work, but you need the Admin PageBlaser Configuration page. This is where your Portal Rules are stored now so I'm surprised you haven't missed it yet. Maybe you are adding all your rules to the global config file.

You may have something out of sync from going through the betas.  You can probably fix it by getting the Uninstall script from the DesktopModules/Snapsis/PageBlaster folder and running that SQL in the Host > SQL window.

After running the SQL script, re-install PageBlaster and you should have the Admin > PageBlaster menu item.

FMSnap,

Please reply directly to the notification from this post so that I'll have the e-mail address to send to.

Ok, I have done so.

hi john,

please send me the latest package too. Before i was on your e-mail list, but i didn´t get any e-mails from you in past weeks.

Thanks jens

l e w a ld [add] lundp. d e

I did the SQL provider - uninstall and re-install on my production system running PB 3.1.1 and the admin page no appwears there.

The Ajax pages also work on under PB 3.1.1 but DO NOT work under PB 3.1.4, so I still need a little help here.

/Dave

In the short term, I am restoring to before I added PB 3.1.4 and iFinity and will go at this a bit more slowly.

Ifinity's Humanfriendly URLs sure make the cache directory structure MUCH deeper. Will this have a performance implication?

/DaveS

With PB 3.1.1 I had to add ListX to the exclude list get get the bottom ListX module on my Listed Homes page to show up. It uses Ajax. The top one doesn't.  With 3.1.4 the entire page was blank.  John did you make any changes in PB 3.1.4 that you would have expected to have caused that behavour?  It seems to me that you may have introduced some unplanned behavour in this area in PB 3.1.4.   I agree that xListing should be added to the exclude list, but the caching behavour should only effect the mdoules on the page using Ajax as I see it.

I am re-cloning and testing PB 3.1.4 and iFinity URL provider on my localhost for a day or two.

 < path pattern="(xListing|\.ashx|LinkClick|rss|install|thumb|wizard|cart|\.asmx)" excludeFrom="All" / > still fails in my clone after the license is put in --- I think that it worked prior.

/DaveS

Hi John

Can you please send me the 3.1.5 version as well.  I did send you an email address to send it to.

Thanks

100% CPU problem still occurs here. The memory seems to be getting released periodically now as opposed to before.

I tried to exclude everything so I could enable them one by one and try to figure out where the problem is but the following rules cause the site to not load with an exception in the PB GetConfig function:

            path pattern=".*" excludeFrom="Caching" /
            path pattern=".*" excludeFrom="Compression" /

I also tried to exclude JSMin, CSSMin, MergeCSS and MergeJS  but they all yield the error

Also it looks liek PB is spawning another process, because now when it goes haywire on the CPU I have to kill two instances of W3wp

The path pattern is the key in the excludePath nodes so you can only have one pattern=".*".
I just ran into this same problem recently myself.
I guess since that area has grown over time I need to add excludeFrom as part of the Key, but for now you can put in one at a time, or change the pattern slightly.

The 100% CPU problem is likely going to be in the Replacements. I saw that you eliminated some of the rules with the XXX technique, were you able to narrow anything down that way?


The only time PageBlaster creates a background thread is on Startup of the application to clean old cache files from the folders. This should still be in the same process though. A new W3wp indicates you have a new app pool, or an app pool that has more than one worker process assigned. If you have more than one worker process assigned to an app pool that is used for DNN you will get strange behavior because DNN caching only allows for one process.

NFX,

I have a guess at what may be happening, although I don't know why it would be ok on PB2 and not PB3.

For your embedded video rules like this:

(afpostbody.*?){youtubevid=(.*?)}

If someone forgot to put in the final curly brace then it would keep searching until it found one, which causes a lot of back-tracking in the Regex engine.

It would be better to change it so it stops if it finds some other characters that would signal that it has gone too far, like a space, newline or a <.

Something like this:

(afpostbody.*?){youtubevid=([^ \n<]+)}

I tried changing the rules but didnt see any difference.

Next I tried to exclude everything like so;

            path pattern=".*|\.001" excludeFrom="CompressionAndCaching" 
            path pattern=".*|\.002" excludeFrom="Replacements" 
            path pattern=".*|\.003" excludeFrom="MergeCSSAndJS" 
            path pattern=".*|\.004" excludeFrom="JSMin" 
            path pattern=".*|\.005" excludeFrom="CSSMin"

Figuring I could re-enable each to find the culprit.

I watched the site and it never stayed at 100% for very long but it was noticably using more CPU and memory than PB2, then I saw the CPU spike up and memory spike up and this happened:

Exception type: OutOfMemoryException
Exception message: Exception of type 'System.OutOfMemoryException' was thrown. 

Request information:

Request URL: http://www.warbeats.com/Default.aspx?TabID=57
Request path: /Default.aspx
User: soboy

Is authenticated: True
Authentication Type: Forms

Thread account name: NT AUTHORITY\NETWORK SERVICE

Thread information:
Thread ID: 
Thread account name: NT AUTHORITY\NETWORK SERVICE
Is impersonating: False

Stack trace: at System.String.GetStringForStringBuilder(String value, Int32 startIndex, Int32 length, Int32 capacity)
at System.Text.StringBuilder.GetNewString(String currentString, Int32 requiredLength)
at System.Text.StringBuilder.Append(String value)
at Snapsis.HttpModules.PageBlaster.RequestFilter.Write(Byte[] buffer, Int32 offset, Int32 count)
at System.Web.HttpWriter.Filter(Boolean finalFiltering)
at System.Web.HttpResponse.FilterOutput()
at System.Web.HttpApplication.CallFilterExecutionStep.System.Web.HttpApplication.IExecutionStep.Execute()
at System.Web.HttpApplication.ExecuteStep(IExecutionStep step, Boolean& completedSynchronously)

After this my w3wp process restarted and the cycle continued (memory grew until the process rebooted)

That page is a DNN Repository module and the downloads all fail - they just send 0 byte files.

Something is eating memory for sure...