背景
集群使用openldap來管理用戶，并與ranger等需要安全認證的組件進行整合。運維同學(xué)在處理客戶問題時，誤使用kill -9而不是systemctl stop slapd將openldap殺掉了，導(dǎo)致啟動不起來，報錯信息
Sep 16 10:59:10 10 slapd[96215]: hdb_db_open: database "dc=emr,dc=cloud,dc=tencent,dc=com": unclean shutdown detected; attempting recovery.
Sep 16 10:59:10 10 slapd[96215]: bdb(dc=emr,dc=cloud,dc=tencent,dc=com): BDB0110 Log sequence error: page LSN 267 4517301; previous LSN 267 6477499
Sep 16 10:59:10 10 slapd[96215]: bdb(dc=emr,dc=cloud,dc=tencent,dc=com): BDB1520 Recovery function for LSN 267 6477715 failed on forward pass
Sep 16 10:59:10 10 slapd[96215]: bdb(dc=emr,dc=cloud,dc=tencent,dc=com): BDB0061 PANIC: Invalid argument
Sep 16 10:59:10 10 slapd[96215]: bdb(dc=emr,dc=cloud,dc=tencent,dc=com): BDB1546 unable to join the environment
Sep 16 10:59:10 10 slapd[96215]: hdb_db_open: database "dc=emr,dc=cloud,dc=tencent,dc=com" cannot be recovered, err -30973. Restore from backup!
Sep 16 10:59:10 10 slapd[96215]: bdb(dc=emr,dc=cloud,dc=tencent,dc=com): BDB1566 txn_checkpoint interface requires an environment configured for the transaction subsystem
Sep 16 10:59:10 10 slapd[96215]: bdb_db_close: database "dc=emr,dc=cloud,dc=tencent,dc=com": txn_checkpoint failed: Invalid argument (22).
Sep 16 10:59:10 10 slapd[96215]: backend_startup_one (type=hdb, suffix="dc=emr,dc=cloud,dc=tencent,dc=com"): bi_db_open failed! (-30973)
Sep 16 10:59:10 10 slapd[96215]: bdb_db_close: database "dc=emr,dc=cloud,dc=tencent,dc=com": alock_close failed
Sep 16 10:59:10 10 slapd[96215]: slapd stopped.
解決方式
客戶集群開啟了openldap的高可用，所以需要從正常節(jié)點把openldap里的條目拿過來進行恢復(fù)，操作步驟如下：
# 正常節(jié)點：
slapcat -n 2 -l /root/ldap_bak.ldif
# 異常節(jié)點：
scp好節(jié)點上的/root/ldap_bak.ldif文件
systemctl stop woodpecker-agent
systemctl stop slapd
# 保持目錄權(quán)限一致
cp -r /var/lib/ldap /var/lib/ldap_bak
rm -rf /var/lib/ldap/*
cp /var/lib/ldap_bak/DB_CONFIG /var/lib/ldap/
slapadd -l /root/ldap_bak.ldif
chown ldap:ldap -R /var/lib/ldap/
systemctl start slapd
systemctl start woodpecker-agent
驗證：
slapcat -n 2 -l /root/ldap_bak_2.ldif
/root/ldap_bak_2.ldif 與 /root/ldap_bak.ldif 對比一致
