OS: Redhat Linux
Submission from: (NULL) (126.96.36.199)
We are using 4 ways multimaster openldap 2.4.11 for more than 5 years.
On Jan 3, 2014, one of the openldap contextCSN is not updated, always stay at
Jan 3 01:xx:xx, all the other servers are already updated to Jan 6.
So, this introduce large amount of sync data between the 4 openldap servers
finally, cause the slapd process memory increase from 500mb to 4Gb and finally,
all memory used up and all 4 servers hanged and error message in KVM shows
"Kernel panic, out of memory". The servers needed to reboot in order to resume
However, after some time, it will happen again.
We finally find that the contextCSN in one server is not updated. What we do is
stop the slapd process in that server, delete the data files and the do a backup
in other 3 servers. Then restore the backup in the problem server from the
backup using slapadd and finally, all the data is synced.
What's the root cause of this problem ? We have been using it for over 5 years
already and this is the first time that this problem occurs.