Re: Waited Too Long For Row Cache Enqueue Lock

If this is an smp (multi-cpu) box, check the value of db_block_lru_latches. Sounds like you may be getting some contention for a latch for the hash buckets that control access to the buffer cache. Basically, a hash bucket represents a "hashed" address for a dba (data block address), and these buckets are divided into groups. Pre-7.3, you could only have one latch to protect the entire group of hash buckets. Under 7.3, with multiple cpu systems, you can have up to 2 times the number of cpus. Unfortunately, the default value is 1/2 the number of cpus, with a minimum of one. Add a line to your initSID.ora file:

db_block_lru_latches = N

where N is twice the number of cpus. Remember that this does nothing for single processor machines. Anyway, without specific error messages and a good bstat/estat run it is pretty hard to do more than guess here.

Roger Snowden
Sr. Systems Engineering Specialist
Oracle Corporation
rsnowden_at_NOTus.oracle.com <-- to reply, remove the obvious

PS- all opinions are mine, personally, and not those of Oracle.

Bill McNamee wrote:
>
> This error message has showed up in our longs a few times over the last few
> weeks. I haven't been able to track down any info on this error. Does
> anybody have more insight into these entries. We are running Oracle 7.3 on
> an HP UNIX box.
>
> Thanks,
> Bill
Received on Mon Jan 11 1999 - 21:32:54 CST