WordPress MU

Oh boy this could be a real show stopper for me :(

I've been playing around with Wordpress MU since before Christmas and i've been extremely happy with. However i've noticed my live testpot domain getting slower and slower and just assumed its my web host at fault.

Today however the site is almost non responsive and taking around 40 to 50 seconds to generate a page.

I'm not live yet, only have 3 test blogs and myself and google to keep me company, have disabled ALL plugins and pretty much most of the MU ones and still having the same issue.

Now what is really bizarre is that my other two sites, one based on Drupal and another thats a flash gaming site, hosted on the same host package (i use addon domains), using the same databases, running from the same domain provider are running lightning. This is where I'm starting to get very concerned.

I've checked my main page and with pretty much everything as standard i'm generating 85 queries which does seem excessive. I think i only generate around that on my main drupal site with around 30 plugins enabled.

I'm running V1.0 release not nightly and really need some help and guidance here. What would people confirm as a standard set of sql queries ? I know its not exactly an exact science but i'd love some feedback on the amount of queries your sites generate.

What concerns me more is, looking at the forums and doing a search for slowdowns, it would appear i'm not at all alone in this finding. I'm still not entirely sure its SQL issues either.

Any help and advice really would be welcome at this stage as i've been developing for a little over a month, love the system but if this happens now, what on earth will happen when i go live ?

Please help where you can with ANY advice and also let me know what version of MU your running, either release or a revision. It may help me track down the problem.

What have you got running in the mu-plugins folder?

Ah problem solved it would seem

Its the WMPU-BAN.PHP plugin pretty much killing the site. Disabled and of it went.

Would still love to know what people consider a reasonable amount of sql queries and what version REV they feel safe using in production sites.

Well, I'm using a bastardized and hacked version of the pre-1.0 release plus a few 1.0 files. :D

Well i assumed i'd solved my problem and everything was working well until i went into the admin control panel. Now when i access plugins it takes like 40 seconds to generate the page :(

Something very screwy going on, especially since i haven't altered anything in days :(

I may re-install incase the databases have got screwed, although i have reverted them back to a good known working version. God knows whats happened here.

By default, aka a clean install, I think the query report is something like 70 or 80.

Which, to me, does seem pretty high initially. Then again, maybe not.

At least thats one thing i haven't to concern myself with :)

I run at about 100 to 120 with everything loaded.

I've had to re-install and that cured my problem however the site export posts feature didn't work so i'll either have to restore my database backup, which could potentially give me issues, or bite the bullet and loose a months work :( I'm not entirely happy at the moment :(

At least i don;t have members shouting from the rafters at me. Thats a blessing.

What concerns me more is how it got to this state as it was functioning perfectly until today.

As i'm doing this, is there a better, more stable release than MU 1.0, maybe one of the REV's i could download that has some of the bug fixes running ?

I did try the REV before this new 848 which worked very well but ignored the categories count number form the options and didn't allow validation on XHTML (hardly show stoppers but also not what i really wanted for production).

Nightly builds are tough to gage. Sometimes their OK, sometimes not.

It's disheartening that you don't have control over your box. I've been tracking what appears to be a random problem, and it spikes the crap out of my httpd process and sends the server into a load of over 30. The only solution I've found is restarting apache, but even ssh takes about 5 or 6 minutes to get into it and get it restarted.

I'm wondering if it would have helped your situation.

What I've been tracking down is definitely somewhere on the MU side, but I haven't figured out what is causing it or why. It's happened probably 4 or 5 times over the past couple of weeks, and it's always been when I was in the backend and doing something with posts.

I haven't been able to get anything off the error log as it hasn't written anything yet by the time I get in and restart apache.

Its a very odd occurance thats for sure.

Anyway i'm pleased to say that i bit the bullet and upgraded to the latest TRUNK and couldn't be happier at the moment. I know its not recommended but i'm still using the site as a test pot and what i can and can't do and having no current users allows me some luxury :) I haven't yet managed to get XAMP to play nice with Wordpress (never can seem to get that darn mod_rewrite to work properly for some reason).

Anyway I updated SVN, deleted everything, exporting the data first to XML files and then re-installed.

When i tried the TRUNK at the weekend it was very buggy, had XHTML errors and pretty much was a show stopper for me.

This latest version however 849, is pretty darn awesome and so far seems very stable. In fact my queries have gone down on the frontpage from 120 to 70 fully loaded with my plugins.

I was really upset when i found the exported XML from MU 1.0 didn't import very well and i had lost data. However the very same XML imported to this version worked perfectly.

Coupled with that, TinyMCE 2.08 out of the box, and the image heading cropping and upload code for themes, i'm in heaven.

Doncha deserves any credit i can give him as he's very much on the ball with updates.

I haven't tested everything yet but this has worked without me having to apply the countless small fixes so far (things like blog creation etc) so again i'm very pleased.

And my themes, the ones i've spent an age getting fully validated and totally error free using Tidy are now once again clean and green. Woohoo.

One thing i did notice is that hooks into TINYMCE from plugins like the contact forms and Imagemanager don't show now but funny enough my modified Advanced WYSIWYG plugin works perfectly. Minor issues in the scale of things really :)

I'm still concerned about why the system got the way it did to be honest and it does have me baffled. It was so smooth yesterday and then went to nowhere land today for no apparent reason.

Like yourself lunabyte i checked the error logs etc and nothing was apparent. I even optimized tables etc but nothing seemed to stick.

I'm running wp-slimstat and thought "ah i bet its that" but alas, disabling it made no difference. In fact i went through a whole bucket of disabling and re-enabling things and never really found a full solution.

One thing that crossed my mind was the actual parsing of plugin names in the Plugins menu. Does that pull data each and every time from each plugin to display that data in the menu or does it cache it somewhere ? The reason i ask is obviously the more you add, the slower the page generation would be unless cached. I'm just curious, anybody know ?

Plugins do read in the files, as do themes as well.

Since it's a file read, it wouldn't be cached unless there was a php caching system in place. I believe APC is the current recommendation.

Nice to know about v849 though. I might look into it and see what's going on.

Its worth it for that custom header implementation alone Lunabyte :)

Still yet to find any major bugs.

I run a local copy that I svn to the nightlies and check it out at home. :)

yes, with XAAMP. :)

I can't seem to get XAMP to play nicely with mod_rewrite Andrea. Any tips ? I've uncomented the module and even moved it around in the conf on some people suggestions but still nothing. Mod_rewrite simply refuses to work for god knows what reason.

I'll have to poke around a bit and see. i know it all works except for the whole subdomain part. *rolls eyes* (mostly haven't bothered to fix it)

Custom header option, huh?

Care to elaborate on that? I haven't checked up on the changes much. Although I'm almost dreading reloading some of the custom edits I've made. Notes I got, time I don't. lol

oh please do as its been a bug bear of mine for sometime :) Sub domains aint an issue as i don't intend to use them.

I cheated at home, and edited my hosts file to test subdomains recently. It's kinda fun really, since you can use any domain you'd like. :D

Custom header option? Nothing different that I haven't seen before, so dunno what you mean. Still just changes the colors.

Correction on previous reply: I have xampplite if that makes a difference. I got it running by following the advice of podz here. Maybe there was something in there.

I think I remember looking in the conf file and setting that up, but that was while installing. I think.

(here, watch my hands, see where I point. :D)

luna, i don't know what could be your problem, but just for the sake of it, you should try a loop on your blogs looking for a category whose parent id is its own. This happens from time to time on my version that a user can do that, and this results in load skyrocketing as you mention. I have a batch running at night fixing all such occurences until i figure where the problem's from.

On another note, MU really generates a lot of queries. 80-120 or such... I've audited them a bit, and the number mostly comes from the habit in WP to get options one by one (would they be site options, blog options, plugin options, user meta date, post meta data etc.) instead of bulk load them and then hit in a cached array. These queries are very fast, since it's only reading in small tables, but yet it brings in the overhead of parsing the SQL, sending the query to the db server, fetching the results etc. Anyway, it's on my todo to rewrite the get_option(s) functions into one that caches everything ;-).

Quenting your dead right about the option calls. I did a small audit on the sql calls myself and was amazed it was pulling these options one by one and as you say this should really be addressed. Granted they are small calls but even so its very inefficient.

Regards the parent id being its own and you running a batch to fix it, would you care to share it ? Sounds like something to implement until we can address that issue. I assume you plough through the sqls looking for that and changing yes ?

Andrea

Custom header option? Nothing different that I haven't seen before, so dunno what you mean. Still just changes the colors.

Oh no no no :) Its the same functionality as wordpress.com. You edit your themes and implement some basic code and bingo, your image headers are now fully changeable (aka the user can upload his own image for the header in a much easier way). It even has cropping facilities so if your header image is 600 x 200 (you define) and a member uploads an image that is 1024 x 700 then a bounding box of the actual image size is positioned around the uploaded image allowing the member to crop his upload to the correct size in the place he or she desires.

It also adds custom sitename and slogan color picking like the default theme and the ability to hide both from each theme too.

The edits are pretty simple but you'll have to update each theme to take advantage.

You can see and find out more about it, with some example edits for themes, by visiting http://boren.nu/archives/2007/01/07/custom-image-header-api/

Thanks for the lammp tutorial but thats exactly what i do and nothing seems to bring in those darn friendly urls. I read that some versions of apache were broke and you need to move that rewrite line up to a certain place, have tried that and still no joy. I'll have another go though as i really could do with a local test pot for the nightly's.

Its bizarre as i've had it working in the past with old phpnukes etc. Dunno what the problem could be.

OooOOOOooooo! *Now* I get it.. Sneaky guys, sneaking it in there like that where I didn't see it...

This'll make my job so, SO much easier.

Yes its great and i plan to update the themes i'm hosting when i can.

Coupled with the template customiser from Jason Ling and you now have the potential of hundreds of variants from one stock theme :)

Dude, we are on the same exact page. :D I was just working on some themes locally with the Jason's customiser. I don't want it as a plugin, so I add the code to the functions files just on the themes I want (and face it, for some themes it doesn't make sense).

But for some reason today, on a live server, it spews out an error over the static command (just a few lines in). *sigh*

Also dizzy, if you carefully read that link above it sends you to a theme repository for wp.com. :)

And just because I was testing: Jason's customiser stuff gets stripped out in the latest nightly of the regular WP. Builds the options page, doesn't do anything with the CSS. So there's the potential for that being some work in the future.

@ Quenting

Nope, nothing out of the ordinary.

I'm wondering if it might not be spam hits on SK2, or direct to my comments file.

although by posting direct, at the very least my captcha should make it die.

It's been very strange, and every time it's happened, there hasn't been more than 100 users on my (primary) site at once. Which isn't squat, so I dunno.

I'm still digging and monitoring though, so we'll see.

Andrea

Also dizzy, if you carefully read that link above it sends you to a theme repository for wp.com. :)

Great eyes :) Shame most of the themes i seem to host arn't there. Guess its back to manually editing them :(

And just because I was testing: Jason's customiser stuff gets stripped out in the latest nightly of the regular WP. Builds the options page, doesn't do anything with the CSS. So there's the potential for that being some work in the future.

Ah nuts yes your right. I just tried it on the last build of MU and nothing is being displayed.

I guess its not fully 2.1 compatible :( One step forward, one step back :)

dizzy, if you could visit my site over here: http://www.hadithuna.com, and tell me if you're able to witness the same kind of lag you're receiving on your own site then perhaps we could try to tackle this issue together.

Right now I have about 180 queries running, which is very high but nonetheless essential to my MU site.

Let me know!

Yes suleiman thats almost exactly what i was seeing.

The latest nightly and enabling the cache has seriously helped though

did anything significant break from the latest nightly? Personally I dread upgrading my MU, as something almost always breaks.

and btw, I've got cache already enabled on my site. :(

Hey suleiman, could it be a bandwidth throughput issue for your box?

Your page creation times are "reporting" 1.x seconds on average, which isn't completely outrageous. But, the page does take a bit to load up visually.