APC CSS and JS check

[davidstrauss]

No description provided.

[catch56]

Also http://drupal.org/project/agrcache stops the file_exists() from executing altogether without a core hack, and solves several other performance issues when css and js aggregates have to be generated.

[joshkoenig]

@chx - A subsequent check should have added a check to insure that apc is enabled. See the current state of master:

https://github.com/pantheon-systems/drops-7/blob/master/includes/common.inc#L4871

@catch56 - We need this check in DROPs because we can't require people use certain contribs and the impact on the average site using a network file system is non-trivial.

[davidstrauss]

@catch56 Having a module that exploits hooks to "patch" core is worse than just patching core. You end up duplicating a bunch of code, obfuscating the actual change, and increasing the chance of missing an upstream bug or security fix.

[davidstrauss]

--enable-apc-pthreadrwlocks solves the APC hang (HT Rasmus)

Sounds like a bug in APC.

[catch56]

There's a core patch at http://drupal.org/node/1014086
On 12 Feb 2013 19:30, "David Strauss" notifications@github.com wrote:

--enable-apc-pthreadrwlocks solves the APC hang (HT Rasmus)

Sounds like a bug in APC.

—
Reply to this email directly or view it on GitHubhttps://github.com//pull/4#issuecomment-13451273.

[btopro]

Is there a reason not to have this patch live on d.o. in a thread so that it can be included in performance optimized drupal distros? currently d.o. packager wouldn't be able to pick this up and if there's no objection to having it on d.o. I was going to open an issue just to get the patch there / increase visibility.

[davidstrauss]

It's a good general-purpose patch, and I know folks outside of Pantheon use it effectively.

[btopro]

Cool @davidstrauss thx for the notice. A coworker and I are thinking of taking the performance enhancements present in ELMSLN and rolling them into a fork of Pressflow for more general usage. After reviewing the Pressflow changes this probably makes more sense then pulling in this 1 enhancement alone :)

[nstielau]

Awesome @btopro, keep us up to date on the progress!

[btopro]

Ok, did some metrics, lots of A/B/C/D comparison and then I published the results to the repo as well.
https://github.com/btopro/presserflow7

https://github.com/btopro/presserflow7/tree/master/_PATCHES -- has a list of all the patches applied. These are all available on Drupal.org. The PATCHES.txt file also includes some of what was used in the settings.php for the APC / "kitchen sink" cache bin / contrib tests.

https://github.com/btopro/presserflow7/tree/master/_RECIPES -- has a drush recipe that can be applied to any site to get it up to kitchen sink status / settings for comparison of the metrics (modules / variables applied).

https://github.com/btopro/presserflow7/tree/master/_METRICS -- xls sheet that has the raw values, some of the methodology, etc.

This was tested against three testbed sites so it should get some independent / additional validation but results were promising; needs more testing but at the very least it's a good series of patches for you all to check out.

[davidstrauss]

Hi, @btopro. Your changes look great. We should consider rolling more of them into Pressflow 7.

Also, while you're welcome to create your own fork, please choose a name more distinct than "Presserflow."

[btopro]

lol yeah that was just for the experiment; sorry. I made it more obvious since this is mostly for getting reaction to try and get some of these rolled into pressflow (where they make sense) and a blog post on the methodology -- https://github.com/btopro/Presser-Flow-FORK

[btopro]

write up -- http://drupal.psu.edu/content/drupal-speed-tuning-analyzing-and-further-optimizing-pressflow