Forums | Mahara Community

Support /
navigation/theme issue after updating to 1.7.2


anonymous profile picture
Account deleted
Posts: 25

14 August 2013, 20:51

I've just updated our production environment from 1.4 to 1.7 and have run into an issue.

As you can see, the navigation is messed up and things arent displaying correctly for the default theme.

I get a ton of error_log things like this on every page load:

EACCELERATOR can't cache: "/var/www/html/mahara/lib/dwoo/mahara/plugins/function.profile_icon_url.php"
EACCELERATOR: Open for write failed for "/var/cache/php-eaccelerator/0/a/eaccelerator-96746.705286": No such file or directory
EACCELERATOR can't cache: "/var/www/html/mahara/lang/en.utf8/group.php"
EACCELERATOR: Open for write failed for "/var/cache/php-eaccelerator/1/a/eaccelerator-91.424564": No such file or directory
EACCELERATOR can't cache: "/mnt/moodle-prd-docs/maharadocs/dwoo/compile/default/var/www/html/mahara/theme/raw/templates/sidebloc
ks/linksandresources.tpl.d17.php"

I've cleared out my eaccelerator cache after the upgrade... is there some kind of permissions issue going on? I didn't experience this issue on my dev server :(

Thanks for any help!

Tony

 

 

Aaron Wells's profile picture
Posts: 896

15 August 2013, 17:20

Hi Tony,

I'm not familiar with eaccelerator, but from what I've read about it just now, it doesn't seem like Mahara 1.7 should be adding anything that prevents it from working. Here at Catalyst we use APC, another PHP opcode cache, and it works just fine with Mahara 1.7, so there's nothing in the code that makes it uncacheable.

I'd recommend asking for advice on an eaccelerator forum, since this seems not to be a Mahara-specific issue. Since you just upgraded, I would guess that it's most likely a file permissions issue, but that's just a wild guess. Smile

Good luck!

Aaron

A post by Account deleted was deleted

anonymous profile picture
Account deleted
Posts: 25

16 August 2013, 11:41

Just as I suspected--

We have 3 load balanced web servers, each of which stored their own e-accelerator cache locally. This always seemed funny to me, even though it worked fine in the past.

I moved the e-accelerator cache to a shared network mount so that all 3 servers share the same cache now and it fixed the issue! Simple enough... whew.

-Tony

4 results