SQL Server 2017 在帶有 Pacemaker 故障轉移的 Ubuntu 上啟動時崩潰
我已經在 Ubuntu Server 16.04 LTS 上安裝了 mssql-server 和 mssql-server-ha。我在兩個節點上使用 drbd,pacemaker 和 corosync 試圖控制兩個節點之間的自動故障轉移。
crm status
顯示2個錯誤:Failed Actions: * res_mssql_monitor_5000 on hostname2 'invalid parameter' (2): call=57, status=complete, exitreason='2017/11/09 12:33:01 Expected local server name to be res_mssql but it was hostname1', last-rc-change='Thu Nov 9 12:33:01 2017', queued=0ms, exec=5241ms * res_mssql_start_0 on hostname2 'unknown error' (1): call=6086, status=complete, exitreason='SQL Server crashed during startup.', last-rc-change='Thu Nov 9 12:32:39 2017', queued=0ms, exec=24329ms
(實際主機名替換為“hostname1 and hostname2”)
TL;DR 如果有人在 Linux 設置上使用浮動 IP 成功配置了雙節點起搏器/corosync/drbd SQL Server 2017,我很想知道我做錯了什麼。如果您需要其他配置或日誌文件,請告訴我。
我不知道它在哪裡找到實際的 hostname1 vs rs_mssql 作為預期的主機名。上面的錯誤出現在主機名 2 上,所以我認為這可能是我在初始設置期間將配置文件從主機名 1 複製到主機名 2 時發生的。
我的CRM配置:
(注:我還沒有攻擊IPaddr2問題;我的正常IP地址有ens160和ens192,稍後我想配置一個IP別名為ip_mssql,用於公共IP訪問SQL伺服器)
node 1: hostname1 \ attributes node 2: hostname2 \ attributes primitive ip_mssql IPaddr2 \ params ip=(virt IP addr) iflabel=ip_mssql \ #I think iflabel is wrong op monitor interval=5s nic=ip_mssql \ meta target-role=Stopped primitive res_drbd_mssql ocf:linbit:drbd \ params drbd_resource=mssql \ op start interval=0 timeout=240s \ op stop interval=0 timeout=120s primitive res_fs_mssqlData Filesystem \ params device="/dev/drbd0" directory="/var/opt/mssql/data" fstype=xfs \ op start interval=0 timeout=60s \ op stop interval=0 timeout=120s primitive res_fs_mssqlLog Filesystem \ params device="/dev/drbd1" directory="/var/opt/mssql/log" fstype=xfs \ op start interval=0 timeout=60s \ op stop interval=0 timeout=120s primitive res_fs_mssqlTempDB Filesystem \ params device="/dev/drbd2" directory="/var/opt/mssql/tempDB" fstype=xfs \ op start interval=0 timeout=60s \ op stop interval=0 timeout=120s primitive res_mssql ocf:mssql:fci \ op monitor interval=5s timeout=30s \ op start interval=0 timeout=60s \ op stop interval=0 timeout=60s group mssqlserver res_fs_mssqlData res_fs_mssqlLog res_fs_mssqlTempDB ip_mssql ms ms_drbd_mssql res_drbd_mssql \ meta notify=true master-max=1 master-node-max=1 clone-max=2 clone-node-max=1 colocation col_mssql_drbd inf: mssqlserver ms_drbd_mssql:Master order ord_mssql inf: ms_drbd_mssql:promote mssqlserver:start property cib-bootstrap-options: \ have-watchdog=false \ dc-version=1.1.14-70404b0 \ cluster-infrastructure=corosync \ cluster-name=mssqlserver \ stonith-enabled=false \ start-failure-is-fatal=false \ last-lrm-refresh=1510177588 \ startup-fencing=true \ enable-startup-probes=true \ symmetric-cluster=true \ stop-orphan-actions=true \ stonith-action=reboot \ remove-after-stop=false \ stop-all-resources=false \ stop-orphan-resources=true \ no-quorum-policy=ignore \ is-managed-default=true
我可以手動啟動
mssql-server
就好了:sudo systemctl start mssql-server sudo systemctl status mssql-server mssql-server.service - Microsoft SQL Server Database Engine Loaded: loaded (/lib/systemd/system/mssql-server.service; enabled; vendor preset: enabled) Active: active (running) since Thu 2017-11-09 12:49:21 CST; 1s ago Docs: https://docs.microsoft.com/en-us/sql/linux Main PID: 3368 (sqlservr) Tasks: 62 Memory: 171.0M CPU: 1.770s CGroup: /system.slice/mssql-server.service 3368 /opt/mssql/bin/sqlservr 3371 /opt/mssql/bin/sqlservr Nov 09 12:49:21 hostname2 systemd[1]: Started Microsoft SQL Server Database Engine.
這些是我發現的唯一實際錯誤
/var/opt/mssql/log/errorlog
:2017-11-09 12:49:28.17 spid4s Service Master Key could not be decrypted using one of its encryptions. See sys.key_encryptions for details. 2017-11-09 12:49:28.17 spid4s An error occurred during Service Master Key initialization. SQLErrorCode=33095, State=8, LastOsError=0. 2017-11-09 12:49:31.14 spid22s The Service Broker endpoint is in disabled or stopped state. 2017-11-09 12:49:31.14 spid22s The Database Mirroring endpoint is in disabled or stopped state. 2017-11-09 12:49:31.17 spid22s Service Broker manager has started. 2017-11-09 12:49:31.37 spid4s Recovery is complete. This is an informational message only. No user action is required.
手動 drbd 故障轉移通過
umount /dev/drbd0 /dev/drbd1 /dev/drbd2
and工作,然後在新的主節點(和 mount…)drbdadm secondary mssql
上反轉該過程。drbdadm primary mssql
我的 /etc/drbd.d/mssql.res conf(/etc/drbd.d/global_common.conf 未從儲存庫更改):
resource mssql { handlers { split-brain "/usr/lib/drbd/notify-split-brain.sh root"; } net { after-sb-0pri discard-least-changes; after-sb-1pri discard-secondary; after-sb-2pri disconnect; } volume 0 { device minor 0; disk /dev/VG-SqlData/LV-SqlData; meta-disk internal; } volume 1 { device minor 1; disk /dev/VG-SqlLogs/LV-SqlLogs; meta-disk internal; } volume 2 { device minor 2; disk /dev/VG-TempDB/LV-TempDB; meta-disk internal; } syncer { rate 35M; verify-alg md5; } on hostname1 { address <ip addr1>:7788; } on hostname2 { address <ip addr2>:7788; } }
嘗試使用
systemd
來啟動服務:crm configure edit res_mssql
編輯配置,使其如下所示:
primitive res_mssql systemd:mssql-server \ op monitor interval=30s timeout=30s \ op start interval=0 timeout=60s \ op stop interval=0 timeout=60s
那應該完成同樣的事情。但是,我想資源代理可能需要一些額外的參數,這可能是讓它按照您嘗試的方式工作所需要的全部。
我建議檢查 RA 資訊,看看您是否可以弄清楚您缺少哪些參數:
crm ra info ocf:mssql:fci