Cause 1: Generally due to the effectiveness of a certain node, resulting in the other node hit rate drop, buffer missing data and go to the database to find,
Causes the database server to collapse in a short time.
Reason 2: Buffering cyclical effects, such as 6 hours, there is a buffer peak every 6 hours, serious even can cause the db to crash.
Can I restart it? Restart DB in the short term and be crushed, but the buffer data more, repeatedly restarted repeatedly, the buffer was rebuilt, the server stable operation.
Solution:
1 expiration as far as possible at night, because the night relative, traffic and concurrency will not be very large, you can run the script to clear the buffer.
2 Dual Backup.
3 random failure, if the buffer period of 6 hours, we can randomly set a 3-9-hour failure, so that the work to share at various points in time.
Memcache Buffer Avalanche Phenomenon