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

Re: slapd memory usage

Good morning, "List"!

I just checked the memory usage again - after the weekend - and meanwhile the slapd's (5) are 68 MB big.

What can I do to reduce the memory usage? Is this really no bug or leak? Is this really normal? What are the slapd's doing?
I'm thinking of restarting the process every two days or so but that seems rather ugly, or?

If I can give you any kind of further information, please tell me.

Best regards



Tino Lange wrote:

Tino Lange wrote:

2) You didn't say anything to the 224 MB memory usage for such a small
DB? Or is this related to not-checkpointing (releasing) the logs?

You didn't say anything about what platform you're running on, how many
threads slapd has configured, etc. As such, there's not enough information
here to make a meaningful comment about memory usage.

It's a RedHat 7.3 Linux SMP system. I'll give you more details about the
configuration tomorrow. I don't have access to the machine from here...

Hi again!

After making the changes with the
checkpoint 10240 60
I wanted to give the slapd some time after restart to monitor its memory usage again, but it's again growing in time.

Here once again some words about the database itself:

- LDAP Server mainly serving as E-Mail Adressbook for Netscape/Mozilla/Outlook Express Clients.
- contains about 250 Entries, wich are
- updated via a script every 20 minutes.
- The Database has indices over
index objectClass eq
index cn,sn,givenName,mail pres,eq,approx,sub

As you can see the database is really small. Just some entries.

The computer on which it runs is:
Linux RedHat 7.3 SMP, 2x700 Mhz PIII, 1GB Memory.

slapd.conf contains no option about the number of slpad-Processes.
At the moment 5 are running.

After starting (yesterday 9.00h) slapd uses about 2.5 MB, one day later it's about 20 MB, now (15.55h) its about 23 MB. It's constantly growing.
The same I observed before - I wrote that after about two weeks we had a 250 MB slapd.

I know it will grow because of caching, inedxing and so on - but THAT much?

We're running openldap-stable-2.1.12 with Berkely DB 4.1.25 here, database type bdb, everything compiled on the machine itself.

Can I help with some more information?
Best regards