[Date Prev][Date Next]
[Chronological]
[Thread]
[Top]
Re: weird performance issue
Jacob Walcik wrote:
You're going down some twisty paths way before it's justified. It is
probably just the DB_CONFIG. Remember that you can't just change the DB
cachesize on the fly, you have to run db_recover before the new settings
will take effect. This is all in the FAQ, every one of you responding to
this thread just needs to read more closely.
i stopped slapd and ran db_recover, ran slapindex, and then started
slapd once again. is that the correct sequence to follow? i'm still
not seeing any sort of performance improvement.
i'm honestly not sure how relevant this is but, if i look at the
output of "db_stat -m", it appears that only about 260KB is being used
for cache space (output trimmed):
Since the BDB default cache size is 256KB it appears that your setting
has not taken effect. And judging from the db_stat output "pages forced
from the cache" this default cache size is too small for your database.
I suggest you reread the FAQ and/or the BDB documentation for
information on how to correctly set the BDB cache size.
259KB 752B Total cache size.
1 Number of caches.
264KB Pool individual cache size.
4 Requested pages mapped into the process' address space.
37781 Requested pages found in the cache (99%).
567 Requested pages not found in the cache.
9 Pages created in the cache.
564 Pages read into the cache.
75 Pages written from the cache to the backing file.
496 Clean pages forced from the cache.
23 Dirty pages forced from the cache.
0 Dirty pages written by trickle-sync thread.
54 Current total page count.
45 Current clean page count.
9 Current dirty page count.
37 Number of hash buckets used for page location.
38924 Total number of times hash chains searched for a page.
5 The longest hash chain searched for a page.
66434 Total number of hash buckets examined for page location.
85106 The number of hash bucket locks granted without waiting.
0 The number of hash bucket locks granted after waiting.
0 The maximum number of times any hash bucket lock was waited for.
2165 The number of region locks granted without waiting.
0 The number of region locks granted after waiting.
639 The number of page allocations.
1292 The number of hash buckets examined during allocations
60 The max number of hash buckets examined for an allocation
519 The number of pages examined during allocations
25 The max number of pages examined for an allocation
--
-- Howard Chu
Chief Architect, Symas Corp. Director, Highland Sun
http://www.symas.com http://highlandsun.com/hyc
Symas: Premier OpenSource Development and Support