Error when doing replication (ITS#1861)

[harry_rueter@gmx.de]

Full_Name: Harry Rüter
Version: 2.0.23
OS: LINUX Kernel 2.4.18
URL: ftp://ftp.openldap.org/incoming/
Submission from: (NULL) (62.104.212.72)


Hi,

there seems to be an error, when using the replication feature.
Replication works fine on my system until bigsized data is to be replicated.
Personally this happens on my system, when data from Netscape Communicator
(roaming access) is to be replicated.

My historydata has a size of about 96 Kb,
(netscape-profile.schema -> 
   attributetype ( 2.16.840.1.113730.3.1.402 NAME 'nsLIData'
   SYNTAX 1.3.6.1.4.1.1466.115.121.1.5 ) 
) 
this exceeds 65535 bytes which is the size of 

SASL_MAX_BUFF_SIZE 

( in ldap-init.h).


When doing the change the master-server stores the 96kb data
without any problem.

The replication-process stops with an error.
I show you the error message as written in my logfile :

----------------snipp--------------------------------
ber_scanf fmt ({iat) ber:
ber_scanf fmt ({a) ber:
ber_scanf fmt (o) ber:
ber_scanf fmt (}}) ber:
do_sasl_bind: dn (uid=ldapreplicator+realm=HRNET.DE) mech GSSAPI
SASL Authorize [conn=1]: "ldapreplicator" as "u:ldapreplicator"
slap_sasl_bind: username="u:ldapreplicator" realm="HRNET.DE" ssf=56
<== slap_sasl_bind: authzdn: "uid=ldapreplicator + realm=HRNET.DE"
send_ldap_sasl: err=0 len=-1
send_ldap_response: msgid=4 tag=97 err=0
ber_flush: 14 bytes to sd 12
<== slap_sasl_bind: rc=0
connection_get(12): got connid=1
connection_read(12): checking for input on id=1
ldap_pvt_sasl_install
ber_get_next
sb_sasl_pkt_length: received illegal packet length of 98916 bytes
sb_sasl_read: failed to decode packet: generic failure
ber_get_next on fd 12 failed errno=5 (Input/output error)
connection_read(12): input error=-2 id=1, closing.
connection_closing: readying conn=1 sd=12 for close
connection_close: conn=1 sd=12
TLS trace: SSL3 alert write:warning:close notify
----------------snipp--------------------------------

What makes the error really problematic is,
that there's no error-message in 
the files in /var/run/openldap-slurp 
(the localstate-dir i used to configure).

Specially the file slurpd.replog shows no error
message, the data seems to be replicated correctly.

I have to emphasize again, that the error occurs ONLY 
in the replication-process.

If you want more information about my configuration
or the logfiles you can contact me naturally.

greetings from germany

Harry Rüter


PS.: My Configuration

Suse-LINUX 7.1
Kernel 2.4.18
openldap 2.0.23 (source, configured,compiled and installed by myself)
cyrus-sasl 1.5.27
kerberos V 1.2.5
(all configured,compiled and installed by myself)