Is anyone experiencing site performance issues as of right now?

[BrianShaw]

Oh no! Say it ain't so! Not the DELETED thread!

What deleted thread???

 

[zsas]

If your site is affected by gibberish in a 300 page Deleted thread, kill it, we will be happy to start another one....

 

[Sirius Glass]

If your site is affected by gibberish in a 300 page Deleted thread, kill it, we will be happy to start another one....

That is not gibberish and it makes more sense than the 300 pages about making Kodachrome in grandma's chamber pot.


Siriusly, the problem will not be the long threads, but database coding problems. I have dealt with these type of software problems for many decades. It is in the database software and not the records.

Steve

 

[zsas]

I know just letting Sean know that quarantining or deleting that thread really wd not matter to us. Indexing, parallelism all fun things in DBA land are what are at play....

 

[ras351]

one more find. this sql query is causing the remaining lag:

Copying to tmp table | SELECT DISTINCT t.threadid,t.title
FROM post AS p
LEFT JOIN thread AS t ON (p.threadid=t.threa |{/quote]

Trying to find out what component in our software is generating this. I think once we can get rid of that things will be a lot better.

Left/Right/Outer joins can be expensive, more so if the tables aren't indexed for the columns being compared. Sometimes things can be improved by changing the query or by creating/managing an alternative table with less data (which can break normal form rules but I won't tell).

 

[Sean]

ok found the bad code in the sidebar. It was related to "My recent posts" block. I have re-enabled the sidebar but disabled that one block for now. Things seem to be running ok. I am now in contact with vBulletin, turning over all of our config files so they can help us optimize the server. The next couple days I'll be focusing on that. To give you an idea of how the large threads+bad code in the sidebar issue affected us:

The above chart is for the last 3 days. You can visually see the massive problem arise and fade away once it was resolved. I certainly learned a lot during this drama. I had never really needed to go into the database console to troubleshoot. I have a much better feel now for drilling down to issues from the server application level.

 

[zsas]

Reading SQL logs is like reading sandscript, thanks for all your 'data gnome catching'. Those 2 min loads were something. Thanks for your efforts, all seems to be zippy and smooth /*knock on wood*/

 

[brucemuir]

running well here Sean
thanks for the due diligence!

 

[Sean]

Still running ok. Here is another image of the performance, last 24hrs (the spike on the left is when we had the dramas), you can see on the right how it has settled and response times are fast.

There could still be a stray monster thread lurking around that will hit us again, but now that I am familiar with tracking them down the impact should be minimal.

 

[Molli]

It's running perfectly here in Victoria, Australia now, Sean. When I saw the post earlier this morning (3-4am-ish) about database issues, I just steered clear. I can't believe how quickly you've got it up and running again. I "broke" my website numerous times when I was coding WordPress plugins and, even with just a few hundred pages to worry about, that was a nightmare. You have my sincerest commiserations. Digging through code of any type really isn't much fun and this site is massive. Thanks for all of your incredibly hard work - it's truly appreciated.

 

[Sean]

Thanks, just glad it's running well again. Things should improve over the next few days even more. I have some server config files in the hands of vBulletin techs. They are going to help me optimize the server config specifically for the forum software. I also have some additional caching to be installed soon which should further speed up page loads

 

[Molli]

As someone who has only just switched from a dial up connection, any page that loads within ten minutes is a winner for me Most web sites, I used to sit and play solitaire while waiting for something to show up. I watched my very first YouTube video online (not downloaded to my hard drive over a two hour period) only three weeks ago.

That admission out there, APUG has always been one of the very few sites that I've enjoyed visiting because page load times are quick (oh that, and the amazingly generous, knowledgeable people who frequent the forum and share their inspiring work!). I lurked for a good year or two before actually signing up and the few hours here and there over the last couple of days where there have been some glitches... well, that's a pretty phenomenal track record, to have it running so smoothly for every other minute of the day for ten years!

 

[tkamiya]

This morning, I am seeing some slowness to "recent post" rendering. The home page will load and the left panel loads in about 20 seconds. Then the right 5 seconds later.

The delay is not the extent of the past but this is new since your last update, so I thought I'd report it to you.

 

[Sirius Glass]

OMG! OMG!

The deleted thread is delete!

 

[Jerevan]

The gallery is a bit slow at the moment, when I click on the thumbnails to show photos. The forums works fine, loading instantly.

 

[Sean]

The gallery is a bit slow at the moment, when I click on the thumbnails to show photos. The forums works fine, loading instantly.

I've found some memory errors this morning to do with some of the server caching. I'm working on getting that resolved then things should speed up in the gallery after that.

 

[Jerevan]

Thank ye!

 

[Sean]

The gallery is a bit slow at the moment, when I click on the thumbnails to show photos. The forums works fine, loading instantly.

Is the gallery better for you now?

 

[flatulent1]

The gallery is working good for me, there's the briefest of pauses before images load, but it's not bad at all here.

 

[Jerevan]

Is the gallery better for you now?

Yep, works a bit better now.

 

[Sean]

hmm seems to be getting sluggish again..

 

[lxdude]

I think the root cause is the A. vs. D. threads. It didn't really start going to hell in a handbasket until the latest one.

 

[lxdude]

Siriusly, the problem will not be the long threads, but database coding problems. I have dealt with these type of software problems for many decades. It is in the database software and not the records.

Steve

True dat. Way back in the dark ages when I set up and ran small business databases on CP/M, with db's spread over numerous hard disks or a multitude of floppies, the software was always the problem. Except once when it was the hardware (bad motherboard) that corrupted records as they were entered. Discs did fail, but tape backups always kept things going. The longest records here are proportionally tiny compared to what churned through my 2.5 (later 4) Mhz 64K RAM machines.