Terrible lag loading posts & threads

[tiburonfirst]

whis, you heard roly - server number 4 has been added, and they will try to find a way to make the social buttons load last. now we have to wait and see if that'll help.

i know somebody questioned the performance of liquidweb and looking at dave's tracert that seems to be a valid point.

 

[whiskey]

His post saying that wasn't clear to me Tibs, so I am still confused as to whats being done.

 

[DaveP]

His post saying that wasn't clear to me Tibs, so I am still confused as to whats being done.

If you look at the last 4 in the tracert, the last is the final connection to ECF in the UK and the previous 3 are in LiquidWeb, which is showing much longer ping times. Those pings are for individual packets. An entire screen update when you click on something in ECF could take twice as long to download all the data it takes to paint that screen. Slower ping times translate to longer delays for each packet that contributes to making up a screen.

It makes sense that a slow server or servers in this case could lead to long page feeds. Add to that general congestion on the web and you sit for a while before you can click again. The load time is dependent on the slowest link in the chain. If the data are standing in line in Lansing, Michigan, no data is getting to your PC or it's slowly moving through the bottleneck.

 

[whiskey]

If you look at the last 4 in the tracert, the last is the final connection to ECF in the UK and the previous 3 are in LiquidWeb, which is showing much longer ping times. Those pings are for individual packets. An entire screen update when you click on something in ECF could take twice as long to download all the data it takes to paint that screen. Slower ping times translate to longer delays for each packet that contributes to making up a screen.

It makes sense that a slow server or servers in this case could lead to long page feeds. Add to that general congestion on the web and you sit for a while before you can click again. The load time is dependent on the slowest link in the chain. If the data are standing in line in Lansing, Michigan, no data is getting to your PC.

I get that Dave...did you get what I posted??? LMAO

 

[DaveP]

I get that Dave...did you get what I posted??? LMAO

Sorry whiskey, I still had Tiburon's comment in mind about the added server when I responded to you. I was musing about the liquidweb servers.

Tiburonfirst wrote: whis, you heard roly - server number 4 has been added, and they will try to find a way to make the social buttons load last. now we have to wait and see if that'll help.

i know somebody questioned the performance of liquidweb and looking at dave's tracert that seems to be a valid point.

Then, you wrote ...

His post saying that wasn't clear to me Tibs, so I am still confused as to whats being done.

I was clarifying something you weren't asking about! Sorry, girl! My bad.

 

[Sgt. Pepper]

For whatever it's worth, I am moving at lightning speed today--right now. Whether it continues, I don't know. If you're not, I'm sorry. I haven't altered my computer, because I don't know how to. Hope the sun is shining on you too.

 

[DaveP]

110 milliseconds (1/10 of a second) ping isn't really that bad. It's not good, but lots of people surf the web daily with pings in the 400ms range. When a server goes into the 1000s to pass packets it's getting serious.

Packet backup is usually associated with high ping rates. Gamers get kicked off game servers if their ping is higher than 250ms because they ruin the experience for other gamers who are sharing the delay with one player's lag.

I'm thinking that we are having congestion backups at times that create the "waiting for" and "loading" prompts in our browsers. Somewhere there are periodic backups that bring a link to it's knees. The techs at ECF can run much more meaningful real time 24 hour load testing to determine who owns the problem. It's a problem that can take time and resources to solve.

Upgrades almost always introduce new undocumented "features" (read design and coding errors) that frustrate users. I always used to wait until I got phone calls and exception reports about upgrades before I jumped in. If you wait, other people get to tear their hair out and you can order the incrementally repaired final version discs and it's a breeze. I never wanted to be the guinea pig in an upgrade of OS and application software. When 4.0 was released, I'd try to wait for 4.0.3.3123 before spending hours loading and applying it. 4.1 was even better if you could stall for that long.

 

[whiskey]

Sorry whiskey, I still had Tiburon's comment in mind about the added server when I responded to you. I was musing about the liquidweb servers.



Then, you wrote ...


I was clarifying something you weren't asking about! Sorry, girl! My bad.

No biggy at all Dave

 

[rolygate]

Some more stuff to play with:

http://pageload.monitis.com/pagespe...-what-do-you-bring-mosh-pit.html&testId=30227

From this, you can see that about half the weight on our pageloads comes from the social media integrations - and about 99% of the issues. We've decided to offer a template that is completely free of all social stuff, to see if it helps people with slow pageloads. That will be the big test, since if it isn't that, we've got a problem (that is invisible at our end).

Another thing is that Facebook have apparently switched to using some IPv6 ports without telling anyone, and this is causing problems for many people integrating their service - we might be able to alleviate that problem, now we know what it is.

All our servers are in Lansing, MI, by the way. IPs are not a reliable way of locating a service unless you know the resource telling you where something is has an up to date database, because blocks of IPs are sold off between countries. Old DBs frequently have the wrong info. Here is one that is usually up to date (and fun to use):

Visual Trace Route Tool - Find, Track, and Map the Route to an IP Address

The social-free template should be online tomorrow

 

[whiskey]

Thanks Roly

 

[CalamityJan]

Hallelujah, thanks for your efforts Roly! Now we just wait and pray....

 

[JerryRM]

Some more stuff to play with:

http://pageload.monitis.com/pagespe...-what-do-you-bring-mosh-pit.html&testId=30227

From this, you can see that about half the weight on our pageloads comes from the social media integrations - and about 99% of the issues. We've decided to offer a template that is completely free of all social stuff, to see if it helps people with slow pageloads. That will be the big test, since if it isn't that, we've got a problem (that is invisible at our end).

Another thing is that Facebook have apparently switched to using some IPv6 ports without telling anyone, and this is causing problems for many people integrating their service - we might be able to alleviate that problem, now we know what it is.

All our servers are in Lansing, MI, by the way. IPs are not a reliable way of locating a service unless you know the resource telling you where something is has an up to date database, because blocks of IPs are sold off between countries. Old DBs frequently have the wrong info. Here is one that is usually up to date (and fun to use):

Visual Trace Route Tool - Find, Track, and Map the Route to an IP Address

The social-free template should be online tomorrow

Chris, all this time, I thought the servers were located in the ECF basement, right next to the dungeon............maybe I shouldn't have disclosed that, it's probably classified information.

Back on topic. A template free of the social stuff sounds good to me. I plan to give it a try.

 

[DaveP]

Interesting trace route site, Roly. It did show Lansing as the final hop. My first try got me a 24 hop route to ECF and the next two were 8 hops. Just goes to show you that the internet is a weird and wonderful creation. AT&T routes everything through San Francisco. My 24 hop route went through Missouri, back to Atlanta, then to Miami and from there to Chicago and then Lansing with multiple short hops in between. It was a spider web.

I'll bookmark the site, though. It's a quick way to find out what I want to know at times.

IPV6 is going to be a headache for those of us who understand and are comfortable with IPV4. It's a whole different world and a strange one at that. I can see network engineers fretting a bit through that change. The IT world will be hybrid for quite a while until it all moves to IPV6. No more convenient mental calculations for a while ... is that address in this network? ... hmmm, let me get my IPV6 IP calculator up on my smart phone.

 

[Liscab]

C:\Documents and Settings\Alex>tracert e-cigarette-forum.com

Tracing route to e-cigarette-forum.com [72.52.242.206]
over a maximum of 30 hops:

1 1 ms 1 ms 1 ms dslrouter [192.168.1.254]
2 12 ms 10 ms 11 ms adsl-xxxxxxxxxxxx.mia.bellsouth.net [xxxxxxxxxxxxxx]
3 12 ms 12 ms 12 ms 70.159.200.65
4 12 ms 11 ms 12 ms 70.159.202.15
5 14 ms 14 ms 12 ms 12.81.8.78
6 10 ms 10 ms 14 ms 12.81.8.65
7 17 ms 16 ms 16 ms 74.175.192.174
8 22 ms 20 ms 22 ms cr82.fldfl.ip.att.net [12.123.153.114]
9 17 ms 16 ms 17 ms fdlfl01jt.ip.att.net [12.122.81.29]
10 25 ms 17 ms 18 ms 208.178.58.145
11 60 ms 60 ms 64 ms GIGLINX.TenGigabitEthernet3-1.ar5.CHI2.gblx.net
[208.49.135.162]
12 69 ms 68 ms 68 ms lw-dc2-core3-te9-1.rtr.liquidweb.com [209.59.157
.224]
13 77 ms 70 ms 69 ms lw-dc2-sec1-dist1-po1.rtr.liquidweb.com [209.59.
157.216]
14 69 ms 69 ms 69 ms host.e-cigarette-forum.com [72.52.242.206]

Trace complete.

if my tracert make sense here you got it

 

[rolygate]

OK, the social-free template is now working: Blackend.

_________________
Go to the bottom left of any forum page
See the spin box there, it probably says Kirsch (the default template/theme/style/skin)
Select Blackend
It is a black template with orange text
_________________

If it works for those who have slow pageloads, we'll offer a version of our main template Kirsch with no social buttons. It may not be to your taste - I can't stand it - but it's just for testing purposes...

On Blackend we have had to leave the Facebook login integration because it is part of vBulletin. If it looks as if it is still problematic, we'll then try to hack the code to lose it from that template. May or may not be possible.

I get 3.5 second pageloads, on any thread page, including the 'Pre-vet' one with nearly 8,000 pages, where I haven't been to that page before, while logged-out, using Firefox, no plugins operating, on phone line broadband, 3,000 miles from the server, in another country. Logged-in, it goes up to 4.5 seconds.

 

[whiskey]

Are we automatically switched to it now or do we have to change something in our settings Roly??

 

[rdsok]

Not roly but...

You'll have to select it from the Theme drop down selector found in the lower left had corner

 

[whiskey]

Thanks Rdok

 

[rolygate]

Yes sorry Whiskey - I altered my post to make a bit more sense

 

[whiskey]

Most pages loading fairly quickly, today...but this thread(see link)still takes over a minute to load

http://www.e-cigarette-forum.com/forum/outside/65725-response-threads-4982.html

Maybe this thread can be split up like the ones that were done a while back, for added speed