mdb_idl_fetch_key: cursor failed: Invalid argument (22)

[Roman Rybalko <openldap@romanr.name>]

I have such error in logs:
null_callback : error code 0x50
syncrepl_entry: rid=001 be_add
logID=502,logID=402,logHost=mail.xxx.xxx.xx,logName=main,cn=test failed (80)
do_syncrepl: rid=001 rc 80 retrying (2 retries left)

and then every 2-3 minutes:
=> mdb_idl_fetch_key: cursor failed: Invalid argument (22)

MDB is running on replication consumer.
I shutted down the replication provider for reindexing.
The consumer was not fully synced at that time.
Than I changed olcDbMaxSize on the replication consumer.
Than I started the provider again.
Seems the error happened first time right after olcDbMaxSize was
increased from 5G to 8G. I've never changed it before.
Sep 30 16:07:01 lintest slapd[518]: conn=1089 op=1 MOD
dn="olcDatabase={1}mdb,cn=config"
Sep 30 16:07:01 lintest slapd[518]: conn=1089 op=1 MOD attr=olcDbMaxSize
Sep 30 16:07:03 lintest slapd[518]: conn=1089 op=1 RESULT tag=103 err=0
text=
Sep 30 16:07:03 lintest slapd[518]: conn=1089 op=2 UNBIND
Sep 30 16:07:03 lintest slapd[518]: conn=1089 fd=12 closed
Sep 30 16:07:03 lintest slapd[518]: => mdb_idl_fetch_key: cursor failed:
Invalid argument (22)

slapd version git a1c2dc6
db size 3.7Gb

What this error is about? Is it serious?
Should I recreate the DB on the consumer? Seems replication is stalled.

-- 
WBR,
Roman Rybalko