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

Re: big replog's problem



--On Monday, December 05, 2005 7:38 PM +0800 Yingbo Qiu <qiuyingbo@gmail.com> wrote:

My openldap slave shutdown for some reason and the replog grows to
400MB. When I found it the slurpd have crashed too. I try restart the
salve and run slurpd to resume the replication. But the slurpd crash
soon and report "malloc of 14 bytes failed". Then I run 'top' to
monitor the slurpd process. It will allocate about 500MBs that reach
the FreeBSD's 512MB hard limit.

I found it crash at about 700,000 line of the replog file. But the
whole file contains 20,000,000 lines.

Is there anyone know howto resolve the problem??

Break it into chunks, and let slurpd process those until it is a reasonable size.

What you really need to do, however, is figure out why you are getting such large replog files in the first place, and if it is because of things like a faulty network, I highly recommend moving to using an alternative replication mechanism like sycnrepl (in OpenLDAP 2.3) so that the slaves play catchup rather than the master waiting for them and generating large replog files.

--Quanah

--
Quanah Gibson-Mount
Principal Software Developer
ITSS/Shared Services
Stanford University
GnuPG Public Key: http://www.stanford.edu/~quanah/pgp.html