[Date Prev][Date Next]
[Chronological]
[Thread]
[Top]
Re: High volume ldap queries causing corruption in bdb backend.
- To: openldap-software@OpenLDAP.org
- Subject: Re: High volume ldap queries causing corruption in bdb backend.
- From: Duncan Brannen <dbb@st-andrews.ac.uk>
- Date: Wed, 19 Nov 2003 11:50:37 +0000
- In-reply-to: <370460874.1063913245@cadabra-dsl.stanford.edu>
- References: <938BDCE792A46245AB40A7A87D7312DD022A9124@GAATLP19W.corp.aglrsc.c om> <00d701c373ab$5b087a70$1e06a8c0@gringo> <200309181103.10230.paj1@humboldt.edu> <370460874.1063913245@cadabra-dsl.stanford.edu>
Dear All,
I'm seeing similar problems to Peter. We've openldap 2.1.22 servers using
Berkeley DB 4.1.25 backends. We've backend-monitor enabled on all the servers
OS Solaris 2.8 on Netra's and Debian Linux running on a Dell 2650
Ldap is used for classroom authentication (via samba) , authentication of
web pages
and for email lookups.
None of the servers are highly loaded CPU or memory wise. We've previously
seen high
slapd cpu usage but that doesn't seem to be a symptom this time.
After some time, weeks, ldapsearch operations crawl to a halt, what was
instantaneous
now takes 10s' of seconds.
I've stopped the database & run db_stat with a variety of options & the
database
seems ok, Lock and Locker numbers are well below the Maximum possible,
Numbers like
56M Total number of locks requested.
56M Total number of locks released.
2915 Total number of lock requests failing because DB_LOCK_NOWAIT was set.
32 Total number of locks not immediately available due to conflicts.
0 Number of deadlocks.
0 Lock timeout value.
0 Number of locks that have timed out.
0 Transaction timeout value.
0 Number of transactions that have timed out.
408KB The size of the lock region..
1557 The number of region locks granted after waiting.
106M The number of region locks granted without waiting.
I'm thinking that 1557 locks granted after waiting isn't too bad out of
106M (These are for 1 lock region: 4)
Hits in the caches are 99% - 100% by the time the server dies. I noticed
stopping the server this time
produced a core file but didn't see anything useful in it.
Any hints for setting ldap logging levels to try and get to the bottom of this
(Or backend-monitor searches to look at to see traffic levels) I've got
logging off for
now.
My DB_CACHE file is just
#
#Set DB CacheSize
#
set_cachesize 0 52428800 0
# 0 Gigabytes 50M 0-> allocate contiguously
Cheers,
Duncan