Closed erthink closed 6 years ago
Seems this problem already fixed in ReOpenLDAP by using biglock in the syncrepl message loop. But currently test019-syncrepl-cascade don't use a biglock feature.
Проблема в гонках при каскадном refreshOnly обновлении содержимого в схеме slap1 => slap2 =>slap3.
slap3 выполняет обновление со slap2, в то время когда сам slap2 еще не завершил свою синхронизацию со slap1:
Варианты решения проблемы:
Seems that this issue has the same reason as the https://github.com/ReOpen/ReOpenLDAP/issues/121
diff server1.flt server3.flt
slapd.1.log
slapd.2.log
slapd.3.log