BUGS: 1.9.4.x Dev

[Giorgio Maone]

@GµårÐïåñ:
I had scripts disabled but netflix.com in my whitelist.
Anyway, your redirect loop seems definitely a cookie issue (could you try disabling secure cookie management?), while the last screenshot hints at a missing CSS file.

[GµårÐïåñ]

@GµårÐïåñ:
I had scripts disabled but netflix.com in my whitelist.
Anyway, your redirect loop seems definitely a cookie issue (could you try disabling secure cookie management?), while the last screenshot hints at a missing CSS file.

Well here is an interesting thing. I logged into Netflix and got the usual refresh to homepage. Disabled automatic cookie management and the page refreshed and HOLLY CRAP it shows MemberHome, what the heck? I am now dying to know what just happened and why it works now disabling automatic cookie management. That's it, it works now just fine, I have tried over and over and with other pages that were failing too and they are all working now.

As for the screenshot, it seemed like a missing css to me too but looking at the source, its referenced and it shows up on ABP's list of blockable elements, so its loading but just not taking effect. It only happened after allowing scripts (after being logged in already) so it might be a host of things conflicting with eachother. At least now we know it was surely a cookie issue, now we need to know why.

[therube]

I still have a sneaking suspicion (which of course means absolutely nothing) that something is still eating memory unnecessarily.

Now having a high memory usage in and of itself does not necessarily indicate a problem & may not be one at all.

But I'm seeing oddities, like when typing into this reply box, where my characters are not being echoed right back (some of the time) & at those times I catch CPU usage shooting up to around oh, say 30%, for no apparent reason.

Are you flushing a buffer <in an expensive manner> every 30 seconds ?
Or does when Mem Usage (sitting ~280MB now, was ~420MB earlier with different system) happens to be high, the flush turns into an expensive proposition?

[GµårÐïåñ]

You are not alone. It may be coincidence and a different issue but I get alot of bogged down performance and hangs and page drops too.

[Tom T.]

I'm getting the same symptom I had before, now with 1.9.4.3: Yahoo Classic Mail pages do a little "dip" momentarily -- drop down on the screen about 10cm, then back up, in a split-second, every time a new page loads. Disable ABE -- pages load instantly and correctly. Toggle ABE = toggle the issue on/off.

[Giorgio Maone]

@GµårÐïåñ:
Have you got any extra ABE rule aside the default?
Regarding the cookie thing, how are your HTTPS/Secure Cookie settings (forced/exceptions) exactly?
I've got "automatic secure cookie management" enabled but I could login just fine, therefore must be something specific to your configuration.

@therube:
Have you got any web page refreshing (or performing AJAX activity) in a background tabs when you observe the "hangs" (BTW, are the characters echoed back in a batch with some delay, or just eaten away)?

@Tom T.:
sorry, I couldn't reproduce either on my everyday Fx 3 profile or a Seamonkey 1.x (Gecko 1.8.x like Fx 2, i.e. potentially ABE-problematic) profile.
Could you try the Standard Diagnostic dance?

[therube]

AJAX?
Wouldn't know. I've had any number of differing pages open at some point.
All closed now except for 6 tabs in this forum. Still sitting just below the 300MB Mem Usage mark.
This is the only Ajax I know. How would I know if a page contained AJAX?

When I had other pages open too, I looked for animated GIFs, Flash, ... whatever I would have thought could be causing these periodic spikes, but saw nothing. So then I started closing down all windows except for pages on this forum. Closing the other windows made no difference.

The characters are echoed back after a delay. They are not eaten.

Now I don't want to give the impression that this happens all the time, but it is happening now (& as it was earlier in this same browsing session & as I'm sure it will until I close this session out).

[GµårÐïåñ]

@GµårÐïåñ:
Have you got any extra ABE rule aside the default?
Regarding the cookie thing, how are your HTTPS/Secure Cookie settings (forced/exceptions) exactly?
I've got "automatic secure cookie management" enabled but I could login just fine, therefore must be something specific to your configuration.

No I do not have any extra ABE rules on this machine that is having trouble. The rules are on a VM copy that I am using for testing and tweaking, since officially we are still beta running ABE.

As for my cookie settings, I had enable automatic secure cookies management checked but nothing in force or ignore list. Again, if you recall I have been testing configurations for myself and others on the test machine and once tweaked and finalized, I will copy over and load but right now nothing on this machine, I went ahead and unchecked that which temporarily fixed my problem. Still curious why though.

[Tom T.]

@Tom T.:
sorry, I couldn't reproduce either on my everyday Fx 3 profile or a Seamonkey 1.x (Gecko 1.8.x like Fx 2, i.e. potentially ABE-problematic) profile.
Could you try the Standard Diagnostic dance?

I just now toggled ABE back on, and the issue vanished. (I'm assuming it toggles on the fly, no browser restart required, like the rest of NS Options, correct?)
Beats the hell out of me, Giorgio. If it ever re-appears, will do extensive diagnostic and reproducing steps. Thanks.

[therube]

(I'm assuming it toggles on the fly, no browser restart required, like the rest of NS Options, correct?)

Very good question. Was going to ask about that myself.


Got me??? This is with SeaMonkey 2.
Yesterday with SeaMonkey 1.1.17 (different machine), even though I had high Mem Usage, I did not have periodic spikes like this.




Maybe I should run FF & see if I can get the same to happen there?
(At the least I'll try to run SeaMonkey 2 on a different machine.)

Now this does not start from the onset, but at some point it begins & then will continue until the session is closed.
Perhaps once Mem Usage reaches a certain point? (Which it seems to do rather quickly. Further Mem Usage does not drop on closing of windows.)

[Giorgio Maone]

Yes, ABE disablement/enablement is immediately effective.

@therube:
If those spikes have a 30 seconds period, they can come from one source only, local rules updates checking.

This task is very lightweight, taking less than 1 millisecond in my tests, but however your mileage may vary according to your system specs, especially disk speed.

Could you try the following and check if the spike disappears?

• Open the Error Console
• Evaluate this code:

  Code: Select all

  top.opener.noscriptOverlay.ns.__parent__.ABEStorage._delay = 360000
  

[therube]

On a different machine now.
FWIW, my SeaMonkey 1.1.17 as it has been sitting since yesterday, with a single window open.

http://i40.tinypic.com/28w0a5f.jpg

On 1.1.17 that returns 360,000 as expected. Won't be able to test the other till this evening.

The other machine is relatively powerful, Intel E4300, 4MB RAM.
(I guess its still a dog compared to many, but its adequate & shouldn't be seeing problems.)
What I'm on now is an AMD Sempron 3000, 512KB RAM.

[Giorgio Maone]

On 1.1.17 that returns 360,000 as expected.

OK, but do the spikes go away?

[therube]

Per my link (picture) above, I'm not seeing any spikes in SeaMonkey 1.1.17.
I'm running SeaMonkey 2 on my AMD now, so I'll open what I normally open & do what I normally do & I'll see how it goes later.

[therube]

Now that didn't take long.
I have any number of windows/tabs open (~8/43) & I'm now duplicating on my AMD w/SeaMonkey 2.
(SeaMonkey 2 being in common, though different builds & <possibly> different extensions installed).



Off to try 360000.

Same results. So what? That points to it not being due to some flushing operation?