Thread

Todd, are you running any extensions? Also, are you seeing this behavior when loading a simple template?

Not running any extensions, and, no, I’m not seeing the same behavior anymore on the public side.

Last night, of course, was a different story. Almost every page was spinning endlessly, but when I turned off caching for each each template directly in the database, and killed off the Sleeper connections, the public pages became accessible once again.

So how is this evening going now that you made the changes?

The control panel is still inaccessible. Spins endlessly, which I suspect is why I’m getting so many sleeper connections to the database.

I haven’t upgraded, yet. I wanted to fix access to the control panel before going through that process.

So here’s the latest. I decided to attempt an upgrade to 1.6.3, but on a backup copy of the live website. I copied the live db to a new database, and then just pointed the config.php in the 1.6.3 folder to this new backup copy of the live database. Unfortunately, I ran update.php only to watch it fall into the same pattern as the EE control panel on the live install. I’m now watching it spin endlessly just like the control panel does, and I can even see the connection to this new database listed in Sleep mode.

Hey, Todd - it’s probably best that we work this out before updating.  It sounds like it may be checking for updates and timing out on that process.  Can you try adding:

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

to your config.php and temporarily remove /system/plugins/pi.magpie.php?  make a backup of both files first, please.  Let me know if that helps.

So things are back to normal. Mysteriously, whatever was preventing us from accessing the control just vanished. We’re still trying to figure out what happened. One last question I have ...

Does the EE control panel use the cache directory in any way?

Todd, nothing in the control panel is cached in any way, not even the queries.

Not even that RSS feed from EllisLab that shows up via the Magpie plugin?

Magpie keeps its own cache, independent of the rest of EE.

Yeah, I’ve seen that magpie cache folder in /ee/system/cache. Okay, so the control panel doesn’t cache anything, but if that RSS feed from you guys is enabled then the control panel effectively does rely on the cache folder, indirectly, because of the plugin.

The reason I ask is because we think our problem had something to do with the webserver not being able to write files to the cache folder last week. It’s an NFS mount, and we know our provider, Joyent, was having some NFS problems last week. 

When we turned off caching for all our templates, all the pages came back, but not the control panel. I’m thinking if there was any reliance on the cache folder, even indirectly through a plugin, that might have been why we couldn’t access the CP.

To be honest, NFS is pretty advanced stuff. I don’t see why it shouldn’t work, but we cannot offer support for advanced server configurations, obviously. Couldn’t you just turn off the feed, or even disable/remove the plugin temporarily to test your hypothesis?

We’re simply trying to figure out what happened. Everything is back to normal, but we believe NFS was the problem, not EE. There’s no way to test this, but knowing the control panel had some reliance on the EE cache folder indirectly via the Magpie plugin gives us more reason to believe the problem was indeed NFS.

I see. That said, are ready to close this thread? Any unresolved technical issues left?

Consider it closed.