performing the search ... it returned everything (all data from the initial ldi) in very short time and stopped on the same place as before. After that it took ~10 minutes to finish...
now, the process is taking 39.5% of memory and 15 - 20 % CPU when doing nothing..
USER PID %CPU %MEM VSZ RSS TTY STAT START TIME COMMAND ldap 10597 15.6 39.5 8500816 4046688 ? Ssl 22:05 2:03 /usr/local/libexec/slapd -h ldap:/// -u ldap
T.
ldap 10597 21.9 39.5 8500816 4046688 ? Ssl 22:05 2:03 /usr/local/libexec/slapd -h ldap:/// -u ldap
On Fri, Sep 11, 2009 at 10:09 PM, Quanah Gibson-Mount quanah@zimbra.comwrote:
--On Friday, September 11, 2009 10:04 PM +0200 Tihomir Culjaga < tculjaga@gmail.com> wrote:
What do you mean by: "And setting no cache or idlcachesize"
you want me to comment this out ?
# cachesize 2500000 # idlcachesize 7500000 cachefree 1000 dncachesize 0
For the moment, yes. See what size the slapd process is in real memory with that setting first, after doing a query across the entire database.
--Quanah
--
Quanah Gibson-Mount Principal Software Engineer Zimbra, Inc
Zimbra :: the leader in open source messaging and collaboration
just one more thing,
when i perform the search again (tried this several times), the CPU usage of slapd is almost nothing 1 - 5% until it stopps returning entries ... than suddenly the CPU goes to 80% - 100% for some time ~1.5 minutes and all remaining entries are returned.
For the moment the memory us keeps stable on 39.5%
All entries are returned within 2m43.999s but there is a gap of about 2 minutes when it stops for a while and returns the final few entries.
T.
On Fri, Sep 11, 2009 at 10:20 PM, Tihomir Culjaga tculjaga@gmail.comwrote:
performing the search ... it returned everything (all data from the initial ldi) in very short time and stopped on the same place as before. After that it took ~10 minutes to finish...
now, the process is taking 39.5% of memory and 15 - 20 % CPU when doing nothing..
USER PID %CPU %MEM VSZ RSS TTY STAT START TIME COMMAND ldap 10597 15.6 39.5 8500816 4046688 ? Ssl 22:05 2:03 /usr/local/libexec/slapd -h ldap:/// -u ldap
T.
ldap 10597 21.9 39.5 8500816 4046688 ? Ssl 22:05 2:03 /usr/local/libexec/slapd -h ldap:/// -u ldap
On Fri, Sep 11, 2009 at 10:09 PM, Quanah Gibson-Mount quanah@zimbra.comwrote:
--On Friday, September 11, 2009 10:04 PM +0200 Tihomir Culjaga < tculjaga@gmail.com> wrote:
What do you mean by: "And setting no cache or idlcachesize"
you want me to comment this out ?
# cachesize 2500000 # idlcachesize 7500000 cachefree 1000 dncachesize 0
For the moment, yes. See what size the slapd process is in real memory with that setting first, after doing a query across the entire database.
--Quanah
--
Quanah Gibson-Mount Principal Software Engineer Zimbra, Inc
Zimbra :: the leader in open source messaging and collaboration
hi Quanah,
hope you are ok... any chance you had some time to check this issue ?
On Fri, Sep 11, 2009 at 10:35 PM, Tihomir Culjaga tculjaga@gmail.comwrote:
just one more thing,
when i perform the search again (tried this several times), the CPU usage of slapd is almost nothing 1 - 5% until it stopps returning entries ... than suddenly the CPU goes to 80% - 100% for some time ~1.5 minutes and all remaining entries are returned.
For the moment the memory us keeps stable on 39.5%
All entries are returned within 2m43.999s but there is a gap of about 2 minutes when it stops for a while and returns the final few entries.
T.
On Fri, Sep 11, 2009 at 10:20 PM, Tihomir Culjaga tculjaga@gmail.comwrote:
performing the search ... it returned everything (all data from the initial ldi) in very short time and stopped on the same place as before. After that it took ~10 minutes to finish...
now, the process is taking 39.5% of memory and 15 - 20 % CPU when doing nothing..
USER PID %CPU %MEM VSZ RSS TTY STAT START TIME COMMAND ldap 10597 15.6 39.5 8500816 4046688 ? Ssl 22:05 2:03 /usr/local/libexec/slapd -h ldap:/// -u ldap
T.
ldap 10597 21.9 39.5 8500816 4046688 ? Ssl 22:05 2:03 /usr/local/libexec/slapd -h ldap:/// -u ldap
On Fri, Sep 11, 2009 at 10:09 PM, Quanah Gibson-Mount quanah@zimbra.comwrote:
--On Friday, September 11, 2009 10:04 PM +0200 Tihomir Culjaga < tculjaga@gmail.com> wrote:
What do you mean by: "And setting no cache or idlcachesize"
you want me to comment this out ?
# cachesize 2500000 # idlcachesize 7500000 cachefree 1000 dncachesize 0
For the moment, yes. See what size the slapd process is in real memory with that setting first, after doing a query across the entire database.
--Quanah
--
Quanah Gibson-Mount Principal Software Engineer Zimbra, Inc
Zimbra :: the leader in open source messaging and collaboration
can anyone help ?
On Mon, Sep 14, 2009 at 7:05 PM, Tihomir Culjaga tculjaga@gmail.com wrote:
hi Quanah,
hope you are ok... any chance you had some time to check this issue ?
On Fri, Sep 11, 2009 at 10:35 PM, Tihomir Culjaga tculjaga@gmail.comwrote:
just one more thing,
when i perform the search again (tried this several times), the CPU usage of slapd is almost nothing 1 - 5% until it stopps returning entries ... than suddenly the CPU goes to 80% - 100% for some time ~1.5 minutes and all remaining entries are returned.
For the moment the memory us keeps stable on 39.5%
All entries are returned within 2m43.999s but there is a gap of about 2 minutes when it stops for a while and returns the final few entries.
T.
On Fri, Sep 11, 2009 at 10:20 PM, Tihomir Culjaga tculjaga@gmail.comwrote:
performing the search ... it returned everything (all data from the initial ldi) in very short time and stopped on the same place as before. After that it took ~10 minutes to finish...
now, the process is taking 39.5% of memory and 15 - 20 % CPU when doing nothing..
USER PID %CPU %MEM VSZ RSS TTY STAT START TIME COMMAND ldap 10597 15.6 39.5 8500816 4046688 ? Ssl 22:05 2:03 /usr/local/libexec/slapd -h ldap:/// -u ldap
T.
ldap 10597 21.9 39.5 8500816 4046688 ? Ssl 22:05 2:03 /usr/local/libexec/slapd -h ldap:/// -u ldap
On Fri, Sep 11, 2009 at 10:09 PM, Quanah Gibson-Mount <quanah@zimbra.com
wrote:
--On Friday, September 11, 2009 10:04 PM +0200 Tihomir Culjaga < tculjaga@gmail.com> wrote:
What do you mean by: "And setting no cache or idlcachesize"
you want me to comment this out ?
# cachesize 2500000 # idlcachesize 7500000 cachefree 1000 dncachesize 0
For the moment, yes. See what size the slapd process is in real memory with that setting first, after doing a query across the entire database.
--Quanah
--
Quanah Gibson-Mount Principal Software Engineer Zimbra, Inc
Zimbra :: the leader in open source messaging and collaboration
--On Tuesday, September 15, 2009 5:22 PM +0200 Tihomir Culjaga tculjaga@gmail.com wrote:
can anyone help ?
Since you reduced the number of your entries to where you DB is only 3.4 GB in size, there is no need to use a 7GB DB cache in DB_CONFIG anymore. I'd suggest you drop it to 4GB, use the unlimited dncachesize setting (dncachesize 0), and then add back in some tuning for cachesize and idlcachesize settings.
You need to run one ldapsearch through the entire database to let it seed the dncache and entry cache, and to pull the data into the BDB DB cache.
Once that is done, do your test.
--Quanah
--
Quanah Gibson-Mount Principal Software Engineer Zimbra, Inc -------------------- Zimbra :: the leader in open source messaging and collaboration
openldap-technical@openldap.org
-
Quanah Gibson-Mount -
Tihomir Culjaga