MODX having a hard time on Rackspace Cloud Sites

I'm using the traditional version, but I'm not placing the files by direct upload. I zip the files, upload them, and unzip them with a shell script (a Perl script running as a Cron job because that's the only way to do it in this environment).

That's probably why I'm having trouble. If there are any further problems of this sort, I'll do a straight upload via SFTP.

If it's just file corruption I'm not worried, reinstalling will fix it. My concern is that it might be some sort of problem occurring while running the setup script.

OK, I was wrong, the problem with listing custom objects isn't a JS issue, and it's not just in production either.

It's directly related to the xPDO::OPT_USE_FLOCK => false setting. With flock off, add-ons can't list objects in the manager (even though they are listed fine on the front end). With flock on, they are listed fine in the manager.

Here's what happened: I re-installed MODX on my production install, and in order to do so I had to turn flock back on. But then I forgot to turn it off, and the listing problem in the manager was fixed. So I thought I had solved the problem, turned flock back off, and the listing grids stopped listing data.

Then I remembered that I had left flock on in my test install, having discovered that package management works anyway. So I turned it off, and sure enough, the listing grids broke.

It's completely consistent and reproducible, I don't even have to refresh the manager page - I have some dropdowns in the Webinex component that list custom objects, and I can turn flock on and they work, turn it off and they list nothing - no clearing the cache, no refreshing the page.

I tested this on a completely different site using 2.2.6 on a Cloud Server (Running PHP 5.4 under CENTOS) and the result is exactly the same - so it has nothing to do with the 2.2.7 revisions, it's just what happens when you turn flock off. You should be able to reproduce it easily - turn flock on, install Articles, make an articles container and a single article - it should be listed in the Articles container - then turn flock off and refresh the manager page for the container - the article should disappear. Turn flock back on, refresh the page, the article should reappear.

The one saving grace here is that even with flock on, this version of MODX seems to work much better. I can reproduce the problems I was seeing before, but only by clearing the cache multiple times and refreshing the page on the front end. With 2.2.6, something would break almost every time I cleared the cache, and it would take several attempts to get it to cache cleanly with no problems. I have no explanation for this, that's just what I'm seeing.

So this is still an improvement. I'll see if I can figure out why these listing grids aren't working with flock off. [ed. note: esnyder last edited this post 13 years, 8 months ago.]

Any updates on this? In particular I'd like to know if you have any insight into the problem with listing processors for extras that I mentioned above. Right now my sites seem to be working great other than that - but every now and then I have to turn flock back on so I can modify listed data.

What's the latest on this issue? Is this the post to be watching or has some announcement been made in another one?

From what I gather, the issue has been somewhat alleviated with the 2.2.7 dev build, but not gone completely. Sadly using Rackspace Cloud Sites for production and Modx 2.x revo is not a good idea. We have 2 Cloud Servers spun up for HTTP and then we're using Cloud Databases for the MySQL instances – This advice came from our sysadmin who only moonlights for us and his day job is at Oracle on the MySQL team.

The 2.2.7DEV released definitely did not fix all the issues. We had to move the sites to a cloud server as well. It is not just the cache corruption, working sites are noticeably slower to respond on Rackspace Cloud Sites than on other hosting environments. What is odd is that some sites never have a problem at all while others crash daily.

Is anything still happening with a fix for this? Almost a month a not heard anything from Opengeek.

Not complaining just wondering if it is still progressing?

@Pyrographics Yeah I can confirm the inconsistencies ... and I think it's what bugs me the most. IN any business application uncertainty is a killer. That's why partly on your advice we made a move... right now we're only moving problem sites... but I'm not deploying new sites on RS Cloudsites. perhaps in some cases it makes sense, but until RS changes something I think we're hosed.

It does pain me that I feel like I'm in the same hosting situation I was in years ago now again. Single point of failure. I know MODx Cloud is out there... but I have some concerns there, not technology wise, just scaling my business on it.. I think I'd go broke

Quote from: pyrographics at Jan 22, 2013, 09:31 AM

The 2.2.7DEV released definitely did not fix all the issues. We had to move the sites to a cloud server as well. It is not just the cache corruption, working sites are noticeably slower to respond on Rackspace Cloud Sites than on other hosting environments. What is odd is that some sites never have a problem at all while others crash daily.

Same here. We have nearly 50 sites on RS Clouds. Only about 6-8 of them have been affected. Most I can get back with the cache-clearing method. 2.2.7 has made a few more stable than they were. I've had to move 3 sites to a Cloud Server because they simply would not come back.

I do appreciate all the work going in to this. I understand the MODx devs are stretched pretty thin since they also support MODx Cloud.

Anyone have an issue of not being able to make or edit a chunk? That's my issue specifically. I've been following this thread - I don't have this issue on one 2.2.7 install but i do on my latest?

yes - this is a new install of modx and yes I cleared the cache manually about 500 times ;-) [ed. note: dylanwood last edited this post 13 years, 8 months ago.]