100% CPU usage, very slow response/timeout

I'm not sure I can give you a proper bug report, but let me create this issue, 
and if you have no idea at all, just close it.

What steps will reproduce the problem?
1. we have a 2 queue SET based queues, and 7 SET based cache
2. the queue is quite small, typically have <10 items (with <10 bytes)
3. the cache is about 800 MB, with typically 100 byte -> 8 kbyte values
4. we do about <1000 queries per second according to the MONITOR command
5. sometimes the CPU usage of Redis is going up to 100%, and staying there for 
30-60s
6. While the CPU usage is around 100%, the response time is very slow, 
sometimes over 30s, sometimes timeouting, the CLI benchmark tool says about 50 
GET queries per sec, while typically it says 45.000
7. I have thought the issue is related to background saving, but not
8. We're using the default factory settings, no virtual memory or other like 
this

redis_version:2.0.2
redis_git_sha1:00000000
redis_git_dirty:0
arch_bits:64
multiplexing_api:epoll
process_id:29610
uptime_in_seconds:6006
uptime_in_days:0
connected_clients:20
connected_slaves:0
blocked_clients:0
used_memory:904343768
used_memory_human:862.45M
changes_since_last_save:715
bgsave_in_progress:0
last_save_time:1294150490
bgrewriteaof_in_progress:0
total_connections_received:18036
total_commands_processed:3992865
expired_keys:0
hash_max_zipmap_entries:64
hash_max_zipmap_value:512
pubsub_channels:0
pubsub_patterns:0
vm_enabled:0
role:master
db0:keys=15,expires=0

It's an Ubuntu Linux, on a 8 core processor machine. Redis' memory usage is 10%.

Original issue reported on code.google.com by barthazi...@gmail.com on 5 Jan 2011 at 1:32

Lachim / redis

100% CPU usage, very slow response/timeout #426