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

Re: Syncrepl: Two step forward, one step back

Quanah Gibson-Mount wrote:
--On Tuesday, November 03, 2009 8:02 AM +0100 Peter Mogensen
Is see output like this:
# slapcat | grep 'dn: o=' | wc -l
# slapcat | grep 'dn: o=' | wc -l
# slapcat | grep 'dn: o=' | wc -l
# slapcat | grep 'dn: o=' | wc -l

Why does server-2 "regret" already replicated objects?

Is this expected behaviour?

btw... I'm running 2.4.17

It seems to be a slapcat problem.
I purged the database on both servers and loaded it from new on server-1. Then I started slapd on both servers and ran:

for ((I=1;I<=20;I++)); do slapcat > out-$I; done
for ((I=1;I<=20;I++)); do wc -l out-$I; done

4755 out-1
4755 out-2
4755 out-3
4755 out-4
4491 out-5
4536 out-6
4847 out-7
4831 out-8
4885 out-9
4885 out-10
0 out-11
5252 out-12
5252 out-13
5252 out-14
5252 out-15
5601 out-16
5628 out-17
5601 out-18
5626 out-19
5920 out-20

run 11 seemed to fail silently.
Notice that run 5 is smaller than run 4.

The problem seem to be entries in the LDIF like this:

objectClass: top
objectClass: NamedObject
objectClass: simpleSecurityObject
uid: rieke
userPassword:: e1NBU0x.....
structuralObjectClass: NamedObject
entryUUID: e46b680e-e5f5-102b-93c9-79162adc1d46
creatorsName: dc=admin,dc=example,dc=com
createTimestamp: 20070823185333Z
entryCSN: 20070823185333.000000Z#000002#000#000000
modifiersName: dc=admin,dc=example,dc=com
modifyTimestamp: 20070823185333Z

With an empty "dn:".
There's lots of them in the output.

Running a batch of 20 slapcats without slapd running gives no errors.