options|alloptions race condition may need mitigating

[OllieJones]

There's a notorious race condition in the management of alloptions. See this. https://core.trac.wordpress.org/ticket/31245#comment:57

Figure out whether any special action is required.

[nickchomey]

Interesting. I was digging in a few months ago trying to find the source of mystery issues that only happened with sqlite object cache (not redis object cache).

Once I discovered that alloptions was involved, I think I gave up because I didn't want to go down that rabbit hole.

I'll try to remember what the exact option was that I was having trouble with, but it probably doesn't matter.

I'll also read through the linked Trac ticket and maybe it'll jog my memory, or reveal a solution.

[OllieJones]

It looks like the best, although imperfect, way to handle this would be to, basically, undo the option-bundling feature of of the alloptions item in the cache, and instead hold a cache entry for each of those options stashed in it. I am pretty sure that's what Till means when he mentions using a redis HASH structure to hold it.

What a kludge on a hack on a botch that would be. I could do it, but it will involve a lot of testing and maybe introduce unintended incompatibilities.

The trac ticket was closed six years ago after the $force parameter made it into the wp_cache_get method.

If this is still an unsolved problem I'd love to hear about it.

[nickchomey]

If memory serves me correctly, #82 might actually be related to this. I don't have time to investigate again right now though