Thread

Never happened before but we’ve got some runaway process or query coming from EE that is making one of EE client sites inaccessible. The process commands are simply “Sleep.” I kill them and they come back.

Couple things to note:

We have other PHP/MySQL applications on this server that are running okay, including phpmyadmin and even a CI application. I can reach static files on the server as well.

We’ve restarted MySQL, upped PHP memory requirements and restarted Apache, deleted everything in the EE cache folder ... and we’ve even restarted the server, but the problems persist. 

We’re on a Joyent Accelerator, and it only runs this one website and the CI application.

We can’t get to the EE admin. Anything powered by EE is inaccessible.

If anyone at EllisLab could help us diagnose this, that would be great.

Our problem is very similar to this one ...

http://ellislab.com/forums/viewthread/67969/

One of our problems is that we can’t even access the EE control panel to tweak debugging settings or turn off caching.

Also, there’s nothing in the various cache folders except page_cache. In that folder, I can’t seem to clear a bunch of the folders out ... all with these .nsf…. files. They’re all 0KB.

Todd, you can manually make changes to the config.php file

$conf['enable_db_caching'] = "n";

EEWiki:// Config Options

Thanks ... I did not see that in the config file, but I’ve put it in. We’re also in the process of turning off caching at the template level directly in the database. We’re running 1.6.0, btw.

Things are not getting any better, but I did figure out I could access the default template group that comes out of the box ... “site.” I always keep the group around for future reference. There’s no content for that group anymore, but you can access some of the templates ... and I learned that I could access some system templates such as Login, Member Registration and even the Member list.

But when it comes to any of the templates for the website or even the control panel, we still can’t get any response.

Can you upgrade to 1.6.3 just so we’re on the same code base?

If you can see the site template group contents, but not the other ones you’ve been working on, you might want to simplify them.

Well, I’ve got the site back. I turned off caching for all the templates directly in the database, then killed off all those “Sleeping” processes. Life was good, again. I’ll definitely have to re-evaluate those templates. Odd, though, b/c the site’s been humming along since last October.

I’m upgrading now, but it’s weird ... the update page says I’m upgrading from 1.6 to 1.6.1, but I uploaded 1.6.3. Is that anything to be concerned about?

I’m upgrading now, but it’s weird ... the update page says I’m upgrading from 1.6 to 1.6.1, but I uploaded 1.6.3. Is that anything to be concerned about?

That’s normal, the upgrade will take you through all the versions until it hits 1.6.3.

Did the upgrade work as expected?

When I saw that message, I did not proceed ... thought better to just check, first.

Keep going, then. 😊

It’s normal- you click the link, it upgrades to 1.6.1- then you’ll get another link to upgrade 1.6.1->1.6.2- etc., until you’re current.  Go ahead and finish it off.

And the nsf files are odd- I googled and it looks like they are lotus notes db format.  Which- I can’t see why they’d be filling up the EE cache folder.  That seems really odd.

I think, in our case, these .nsf files are actually network files. The /home directories at Joyent are on network storage servers. I’m working the Joyent end on those, but there was definitely an issue with caching and our templates. And we’re obviously not sure what triggered the problem.

Our control panel is still inaccessible, though I’ve not done the upgrade, yet. I want try to fix access to the CP before going through the upgrade process, especially since the site is working fine on the public side. At the very least, I’ll wait until off-hours before I go through the process.

It just spins and spins, and shows a “Redirect” page title. I’ll send you the URL. If you have any suggestions, send ‘em my way.

Yep- I tried getting it to load with index.php and index.html- nothing but spinning.  I almost wonder if something is up with the server.  But- the update.php came up ok for you- accessing it via same url, just with update.php added?  Well- yes, because I just checked.

Hm- try editing system/index.php to force php errors to show:

error_reporting(0);
change to
error_reporting(E_ALL);

Not a bad idea, though I don’t think that’s what’s going on.

Tried the E_ALL reporting, but that didn’t do anything. Browser still spins and spins endlessly on the control panel. I’ve noticed also Sleeper processes once again building up in the EE database ... which seems to make sense when you think about how the control panel acts. There’s a connection made that just never ends and never results in anything.

Are there any tables I may need to investigate and maybe clear out?