[Date Prev][Date Next] [Chronological] [Thread] [Top]

ldap is too slow, even uncommunicable.



	Im having some problems with OpenLDAP, its getting really slow under
load, although it wasn't. the problem follows: from yesterday to today ldap
seems to stop answerings... im use it as backend for
samba/postfix/pop3/squid so it was working perfect:

Here I've raised the loglevel to 256 to see if it could search, yes it could
search, I could see a lot of searchs going on... but not my search that were
taking too long to show up, in fact they didn't.

When I do ldapsearch -x -z 0 -l 0 , I get:

# search result
search: 2
result: 4 Size limit exceeded

# numResponses: 501
# numEntries: 500

Although I got 583 entries, I checked this using:
ldap:/etc/ldap# slapcat |grep ^dn|wc -l
    583

Hey that's not much.

Basically when I put my samba servers (I got 2) it start getting slow slow
slow ... until I get cant connect to ldap server.

Sep  9 10:03:27 smb smbd[2188]:   failed to bind to server with dn=
cn=admin, dc=tcm, dc=pa, dc=gov, dc=br Error: Can't contact LDAP server

Or either with ldapsearch: Can't connect to LDAP Server.

Here follows my slapd.conf (important sections):

# Read slapd.conf(5) for possible values
#loglevel        0
loglevel        256

# idle timeout
#idletimeout 120
idletimeout 300

# threads
threads 16

im using database ldbm

#cachesize 100000
#dbcachesize 1000000
cachesize 200000
dbcachesize 2000000


rest is default like access lists, index, etc, im using slapd
2.0.23-6.3  , from Debian, 3.0r2 (stable)

Also I get like 20 established connections to my LDAP server and around 60
with either SYN_RECV/CLOSE_WAIT on netstat , all by samba server.


Anyone got any clue where to starting looking for? I already restart ldap
server, recreated indexes, increased its size, also when I put samba servers
down it starts going fast, as its normal operation.

Ah, here are the sizes of my base: pretty small imho.

ldap:/var/lib/ldap# du -sh *
272k cn.dbb
196k dn2id.dbb
2.0M id2entry.dbb
272k mail.dbb
256k mailAlternateAddress.dbb
8.0k nextid.dbb
40k  objectClass.dbb
0    replog
0    replog.lock
132k sn.dbb
268k uid.dbb


Im running a p4 2.8Ghz Xeon, 1Gb DDR , it's a IBM xSeries 345 server.


Thanks on any reply, everyone is stopped because they cant work :( (its
around 450 people)