That was because I was trying - yet again - to find a keyboard method to dismiss the “502 error” message, and it seems I managed instead to tab to the post box, even though it was greyed out on the screen, and instead activated the “reply” button. Believe me, nobody was more confused than I am.
1 Like
Most of these messages respond to Enter, but for some reason, this one doesn’t.
It’s a demon message from coding hell for keyboarders 
1 Like
@eviltrout just fixed a possibly related issue
These 502s are getting so annoying I decided to post with my dev tool network running
The problem seems to be
http://www.sitepoint.com/community/t/146659/posts.json?post_ids%5B%5D=902061&_=1429413880561
Hmmm. there’s json again. Maybe related to my localhost master error?
[vagrant@precise32:/vagrant (master)]$ bundle exec rake db:migrate
== 20150325190959 CreateQueuedPosts: migrating ================================
-- create_table(:queued_posts, {:force=>true})
rake aborted!
StandardError: An error has occurred, this and all later migrations canceled:
PG::UndefinedObject: ERROR: type "json" does not exist
LINE 1: ...ger NOT NULL, "raw" text NOT NULL, "post_options" json NOT N...
^
: CREATE TABLE "queued_posts" ("id" serial primary key, "queue" character
varying(255) NOT NULL, "state" integer NOT NULL, "user_id" integer NOT NULL, "raw"
text NOT NULL, "post_options" json NOT NULL, "topic_id" integer, "approved_by_id"
integer, "approved_at" timestamp, "rejected_by_id" integer, "rejected_at"
timestamp, "created_at" timestamp, "updated_at" timestamp)
/usr/local/rvm/gems/ruby-2.0.0-p0-turbo/gems/rack-mini-profiler-0.9.3/lib/patches/db/pg.rb:90:in `exec'
/usr/local/rvm/gems/ruby-2.0.0-p0-turbo/gems/rack-mini-profiler-0.9.3/lib/patches/db/pg.rb:90:in `async_exec'
/usr/local/rvm/gems/ruby-2.0.0-p0-turbo/gems/activerecord-4.1.9/lib/active_record/connection_adapters/postgresql/database_statements.rb:128:in `block in execute'
I upgraded my Postgres version and switched to beta and now I’m back in business
Just to give everyone an update. The Discourse team did identify a problem and have put in a few patches to the Discourse Core code to help remedy the situation. They’ve released a beta containing those patches today and SP will be looking at integrating that beta.
I don’t have a timeline, nor do I have a guess at one, but it is being actively worked on. This ended up being a bit more complex than what initially met the eye and we had a few moving targets being pegged as the problem early on with these errors.
3 Likes
I haven’t had a 502 error in the past 2 days. Almost afraid to jinx it.
Likewise, though I’ve not posted much at all over the weekend. Fingers-x’d
This should be fixed. We’ve upgraded to beta6 which has a lot of performance improvements (which were part of the 502 issues).
Hmm… 
Scott
LOL
A lot can change in a week’s time, especially when multtiple problems are involved.
Preface “it has” with “to the best of our knowledge” and it makes sense.
In any case - knock wood (they were extremely aggravating) - I haven’t had any 502s for a while, have you or anyone else?
That is the main, most important thing. The problem gets solved.
It was just a bit of fun poking that bit at cpradio.
Scott
Yep, well deserved too! And I definitely don’t mind it
To be fair though, we had two problems. 1) Custom GA code not handling scenarios where category isn’t available, and 2) a performance issue with the number of topics being watched/tracked here.
Both resulted in 502s, the Custom GA code ended up not being the bigger 502 producer of the two. Technically there are now 500 errors happening, but they are very sporadic and I’ve only had it happen once (and I can’t reproduce it at will). The Discourse Team will be investigating our instance further to see if there are other optimizations needed.
1 Like
This topic was automatically closed 91 days after the last reply. New replies are no longer allowed.