[wp-trac] [WordPress Trac] #25623: The design of update_option etc makes race conditions likely to happen

[WordPress Trac]

#25623: The design of update_option etc makes race conditions likely to happen
-------------------------------------------+-----------------------------
 Reporter:  joehoyle                       |       Owner:
     Type:  defect (bug)                   |      Status:  new
 Priority:  normal                         |   Milestone:  Future Release
Component:  Cache API                      |     Version:
 Severity:  normal                         |  Resolution:
 Keywords:  reporter-feedback needs-patch  |     Focuses:
-------------------------------------------+-----------------------------

Comment (by sphoid):

 I'm fairly certain I'm actually having this very problem. We have 7 web
 servers in a production environment using the memcached plugin. I've been
 thinking for the longest time that there is a bug in php-memcache or in
 memcached itself because we intermittently get stubborn stale cache data
 but it sounds very likely that cache writes are stepping all over each
 other causing old data to get written back to the cache which resets the
 expiry for the key. This is a very difficult thing to prove because it
 only seems to happen in high load environments with many servers (this is
 not a common wp configuration outside of wordpress.com) and I'm also
 thinking the size of the data blob might be a factor too since we only
 ever experience this with alloptions which is upwards of 500Kb in our
 environment. I'm watching this ticket with great interest.

--
Ticket URL: <https://core.trac.wordpress.org/ticket/25623#comment:4>
WordPress Trac <https://core.trac.wordpress.org/>
WordPress publishing platform