3.0.4 speed vs 3.0.3

[Connor T]

When I upgraded to 3.0.3 I noticed a horrible change in loading times. I tried a couple of tickets, but nothing got resolved. The staff just kept telling me to disable all my hooks and add-ons till it was the stock software.

Now after upgrading to 3.0.4, the loading times are amazing. In what I tried to tell them in tons of tickets was right. It was IPB 3.0.3 that caused the problems, because 3.0.4 made my board speed up incredibly.

Thanks guys, love the speed now :)

[MageUK]

Meh - I think experience differs greatly for people. With a search I'm sure you can find people who thought it was horrible on 3.0.2 and 3.0.3 fixed that for them, it's the same with every version upgrade. Personally, my main board is still on 3.0.3 and it's extremely fast, In fact I would be extremely impressed if 3.0.4 sped it up at all - I would be happy for it to remain the same. As I said mixed views - 3.0.3 is perfect for me, clearly wasn't for you (and I have a few hooks installed too!).

[ikillbill]

actually, IPB 1.2 is the fasttest, but XD

[Mark]

Matt did loads of optimisation for 3.0.4. Glad to hear it's improved things for you :)

[bfarber]

It's going to differ between users, yes. We've identified many hooks/modifications that cause problems, so naturally if you have one of those you are going to have a different experience from a user that doesn't.

[Michael]

[quote name='bfarber' date='07 October 2009 - 09:26 AM' timestamp='1254921982' post='1864232']
It's going to differ between users, yes. We've identified many hooks/modifications that cause problems, so naturally if you have one of those you are going to have a different experience from a user that doesn't.

Any chance IPS can communicate that list back to the authors of those mods? I would definitely like to know if one of my mods causes problems that it takes a ticket to IPS to resolve. The only one I know about is my Board Panel has a nasty query in it to find the favorite forum, but that does have a setting to disable it should that be a problem.

[bfarber]

We will probably at some point communicate the modifications back to the authors, yes.

[Charles]

We are making a list of mods with known issues and we will definitely be contacting mod authors to see if they can optimize the mods.

[rct2·com]

@ibotPeaches: Whether something is faster or slower is surely a 'perception' rather than a firm statement? Unless you have a 'lab' type server that runs exactly the same transactions on the same database with a known load on the server both times, it's not really possible to accurately judge whether one version is faster than another.

Although it is nice to get confirmation that there has been some optimisation.

[rct2·com]

[quote name='Charles' date='07 October 2009 - 02:47 PM' timestamp='1254923251' post='1864248']
We are making a list of mods with known issues and we will definitely be contacting mod authors to see if they can optimize the mods.


You guys have made it far too easy for people to make mods. Now anybody with low skills can produce something. You should be ashamed of yourselves.

I'm joking of course. :)

[capbiker]

It might be a good idea, maybe after the authors has been notified of problems with there hooks/mods. To put out some kind of info post showing other moder's sections of which was coursing problems and how to fix it for future reference and also how to improve our coding for the future.

cap

[Charles]

[quote name='capbiker' date='07 October 2009 - 10:34 AM' timestamp='1254926055' post='1864257']
It might be a good idea, maybe after the authors has been notified of problems with there hooks/mods. To put out some kind of info post showing other moder's sections of which was coursing problems and how to fix it for future reference and also how to improve our coding for the future.

cap


We were going to post our list but we want to contact the authors first. Don't want to make people look bad for a simple little bug :)

[rbiss]

Do you test mods before you approve them on the resources site? What is the approval process for?

[AndyF]

[quote name='rbiss' date='07 October 2009 - 03:54 PM' timestamp='1254927255' post='1864269']
Do you test mods before you approve them on the resources site?

Yes, new submissions have to be approved before they become public.

[quote name='rbiss' date='07 October 2009 - 03:54 PM' timestamp='1254927255' post='1864269']
What is the approval process for?

To ensure they are safe (ie: do not contain query to drop a table or two or possibly delete (unlink) files), do not contain anything malicious, and are what they claim to be.

There is a bit more to it than that, but that's a brief outline of what I will look for.

:)

[bfarber]

We do not expect our volunteers (like Andy) to go through a rigorous performance testing routine for each modification submitted. To put it bluntly - it's not his responsibility. Andy does a good job just looking through the modifications to try to spot any obviously "bad code" (like he said).

At the end of the day we don't endorse or support modifications, so any time you use one it's at your own risk.

[rbiss]

Do you test IP upgrades against a test board with approved mods?

[Jan Krohn]

I've noticed significant increase in speed once I moved from 3.0.2 to 3.0.3. It really can be faster???
Runs to upgrade..

I've just noticed that this discussion went in hooks direction, so my comment is irrelevant :)

[bfarber]

[quote name='rbiss' date='07 October 2009 - 11:32 AM' timestamp='1254929563' post='1864281']
Do you test IP upgrades against a test board with approved mods?


No. There are no "approved mods" to test against. We do not support third party modifications (or hooks/applications) in any way, shape or form. The closest we come to "supporting" third party code is with our community projects, where we are available for the developers to communicate with if they have problems, questions or suggestions.

[Brandon D]

It's probably Links :x I never went back and profiled it or even take a quick look at the performance

I haven't heard of any problems tho, so let's go with the whole out of sight out of mind thing

[rbiss]

[quote name='bfarber' date='07 October 2009 - 08:59 AM' timestamp='1254931146' post='1864287']
No. There are no "approved mods" to test against. We do not support third party modifications (or hooks/applications) in any way, shape or form. The closest we come to "supporting" third party code is with our community projects, where we are available for the developers to communicate with if they have problems, questions or suggestions.


Given most issues are upgrade problems and possibly mods I suppose the logical step would be stop upgrading or making any changes and thus not renewing.

[nabludatel]

before tell "it can't be faster" show me at least one board running on ipb 3 which is as fast as 2.6 versions. For me it stills a little bit slow, and everybody who was on ipb 1, 2 or other forums (vbulletin, phpbb...) will be agree with me.

[Matt]

I did a fair amount of optimizing in 3.0.4. A lot was SQL side. I removed a lot of temporary tables / filesort operations in fairly common areas (profile, stats, etc). It's the kind of thing a board running well inside its resource footprint would never notice, but a board with a lot of traffic would. For busy boards, the additional I/O and delay caused by MySQL writing temporary files could lock up MySQL which will knock on to all subsequent requests for access.

With regards to third party mods and hooks; as mentioned previously we have been compiling a list and I will email the authors at some point. In most cases, the problem is a slow MySQL query. I can think of one hook off the top of my head that adds a large GROUP by query on the post table on the board index (highly trafficked area). This query is complex and slow causing 1 temporary table and 1 file sort.

There are additional tools in 3.0.4 to help capture slow and 'bad' queries which I'll detail in a resources article.

[ikillbill]

IPB is getting slower

from our experiences 1.2> 2.0.* > 2.1.* > 2.3

People blamed our server, but when we put dirrent version on SAME sever, people then did notice differences

Wish 3.0.* could be tweaked further

i rememebered the final "so -called" conclusion last time is "more function, slower"

So, snice then we decide to only upgrade to new version of IPB "one year" after its release until all major bugs and performance issues resolved

[backdream]

[quote name='ikillbill' date='13 October 2009 - 09:11 AM' timestamp='1255396264' post='1865931']
IPB is getting slower

from our experiences 1.2> 2.0.* > 2.1.* > 2.3

People blamed our server, but when we put dirrent version on SAME sever, people then did notice differences

Wish 3.0.* could be tweaked further

i rememebered the final "so -called" conclusion last time is "more function, slower"

So, snice then we decide to only upgrade to new version of IPB "one year" after its release until all major bugs and performance issues resolved



I agreed it. 3.0 is more slow vs. the 2.x, from browser to server. I hope they can resolve it in 3.1

[AlexJ]

[quote name='Charles' date='07 October 2009 - 03:40 PM' timestamp='1254926424' post='1864264']
We were going to post our list but we want to contact the authors first. Don't want to make people look bad for a simple little bug :)


Understood. But if you guys will publish the list before contacting author I personally feel its more good.

Reason:
1. I/we/all are thankful to that author but since his hook is giving problem we can disable it temporarily till author fixes it.

2. For those who think or will think that those author sucks then they should quit because each software has bugs. Every single thing needs bug fixes.

I hope you guys can release the list because I am using 3.0.3 since their are 2 major issues in 3.0.4 and I am waiting badly for 3.0.5.

Thanks. :)