"unexpected eof" in sui causes it to spin out of control

[erikh]

Not sure what is causing this yet; I'm pretty sure something that's handling errors isn't doing it right, though.

[erikh]

this is related to database size somehow. I am baffled.

[erikh]

Something is causing the database size to grow at a phenomenal pace. One thing that might be related is the recurring primary key which even after the database is rebuilt grows at an astronomically fast pace, starting indexes on my personal account are around 30k (and they should be e.g. 1) so it tells me that there's an inner loop that is getting run without resulting in anything. This could be leading to a buildup of garbage in the DB that causes this issue.

[erikh]

CPU usage during this period looks to be 2-3 cores worth of usage until the error is cleared. After that it subsides, but the database is huge. I'll look into it this weekend, but I'm fairly sure this is the primary key allocator doing something stupid.

[erikh]

Ok, I think I have this dialed into when I modify things in google calendar directly (like moving an entry on my phone), and saturn or sui has to pick them up later.

[erikh]

Actually, this is when sui cannot fulfill the api request. yay.

[erikh]

After a lot more experimentation I still have not nailed this down. Best I can tell something in ciborium and the database size is causing it to partially load the db, which needs more investigation. Slating this for 0.5.0 because I want to get 0.4.4 out the door.