Actually, after going for 2.2.x upgrade, sites are not easily migrating to new platforms.
After solving some minor glitches (most on my part) the strange thing is happening with CSS aggregation: first home page load fails missing aggregated CSS and JS and then subsequent pages (after drush cc all) does have them or they are gradually gaining them.
I cleared the cache on browser and multiple drush cc all, even drush rr but home page simply refuses to recover. That was not happening in prior versions, 2.1.3 and earlier.
Some hint?
Comments
Comment #1
omega8cc commentedThere is nothing in BOA-2.2.3 which could cause this, but the issue is pretty common. Just make sure you have rebuilt AdvAgg cache if used, and otherwise follow standard hints the docs: https://omega8.cc/how-to-disable-all-caching-and-aggregation-115
Comment #2
omega8cc commentedCheck also: https://github.com/omega8cc/boa/commit/db18d4d8d43c539a2893e69f1589ab9b4...
Comment #3
macmladen commentedI know there is nothing specific in BOA 2.2.3 but it just wasn't happening before so I'm noting it as an information, you said yourself it does happen. And I'm pretty sure it is cache question but it is annoying when someone does transition and site opens without CSS (visible in Firebug, clearly without some CSS).
I'm not using AdvAgg myself.
The truth is (it seems to me) that
drushis not clearing the cache while from site itself this clearing solves the problem. It still leaves some CSS missing so first I clean withdrush cc allthen from site clear cache and once moredrush cc all.This seems to solve things with CSS (so far).
Comment #4
omega8cc commentedHow about the clear cache task in Aegir, followed by Verify task? It should do the same, in case wasn't able to clear caches on the fly for any reason. We have seen this happening to our own site in the past, in some very old BOA release, depending on some other factors, like fake-CDN settings etc, so this is something which just may happen under some circumstances, but is not linked to any particular BOA version, of course unless someone will find a way to reproduce the problem consistently and find that it affect any site, so is not codebase specific. Feel free to re-open this issue if you could somehow reproduce this. Just make sure it is not just Redis related, which may happen when you rename the site back and forth, which may leave leftovers in the Redis cache, so temporarily disabling Redis (for 24 hours), may also help.