测试环境:
数据库节点系统:rhel 6.8 x86_64
数据库版本:Oracle 11.2.0.4 RAC
背景:
对比分布式存储和IBM传统存储性能对比。
共三块磁盘,冗余模式采取normal。
在安装数据库集群时,节点一执行root.sh执行失败,提示ORA-27061: waiting for async I/Os failed with root.sh
初步怀疑由于磁盘属性或数组问题,对比两台数据库主机映射磁盘正常,权限正确。
使用grid用户读写共享盘,验证通过
dd if=/dev/urandom of=/dev/asm/asm-diskb bs=4096 count=1000
dd if= =/dev/asm/asm-diskb of=/dev/zero bs=4096 count=1000
多次搜索无果,打算采取外部冗余,测试每块磁盘,此时分布式管理人员该问题可能由于三块共享盘sector size不一致导致。
检查分布式磁盘,发现一块磁盘sector size 是4096,其他两台为512。
[root@htdb01 ~]# fdisk -l
......
Disk /dev/mapper/c1p1: 1600.3 GB, 1600321314816 bytes
255 heads, 63 sectors/track, 194561 cylinders
Units = cylinders of 16065 * 512 = 8225280 bytes
Sector size (logical/physical): 512 bytes / 512 bytes
I/O size (minimum/optimal): 4096 bytes / 524288 bytes
Disk identifier: 0x00000000
Disk /dev/mapper/c2p1: 1600.3 GB, 1600321314816 bytes
255 heads, 63 sectors/track, 194561 cylinders
Units = cylinders of 16065 * 512 = 8225280 bytes
Sector size (logical/physical): 512 bytes / 512 bytes
I/O size (minimum/optimal): 4096 bytes / 524288 bytes
Disk identifier: 0x00000000
Disk /dev/mapper/c3p1: 1600.3 GB, 1600321314816 bytes
255 heads, 63 sectors/track, 194561 cylinders
Units = cylinders of 16065 * 512 = 8225280 bytes
Sector size (logical/physical): 4096 bytes / 4096 bytes
I/O size (minimum/optimal): 4096 bytes / 524288 bytes
Disk identifier: 0x00000000
确认问题后,调整分布式系统上磁盘属性,涉及命令为hdm,软件为:hdm-core-3.3.0-25.ga-x86_64.rpm
安装后检查分布式磁盘信息:
[root@yxcel03 soft]# hdm scan
[1C58SDM00001F5A7HUSMR7632BHP3010023]
Product Name = Ultrastar SN260
Device Type = NVMe Controller
Device Path = /dev/nvme0
UID = 1C58SDM00001F5A7HUSMR7632BHP3010023
Alias = @nvme0
Vendor Name = HGST
Model Name = HUSMR7632BHP301
Vendor ID = 1C58
Device ID = 0023
NVM Subsystem UID = 1C58SDM00001F5A7HUSMR7632BHP301
[000CCA0B018E1280]
Device Type = NVMe Namespace
Device Path = /dev/nvme0n1
UID = 000CCA0B018E1280
Alias = @nvmens0
Parent Product Name = Ultrastar SN260
Parent Device Type = NVMe Controller
Parent Device Path = /dev/nvme0
Parent UID = 1C58SDM00001F5A7HUSMR7632BHP3010023
NVM Subsystem UID = 1C58SDM00001F5A7HUSMR7632BHP301
Namespace ID = 1
......
调整磁盘sector大小
[root@htcel02 soft]# hdm format --sector-size 512 -p /dev/nvme0
This operation will erase all user data on the specified devices. Before starting this operation, unmount any mounted file
systems on the devices or take the devices offline. After this operation is complete, the specified devices will be reset,
which may cause other operations to be interrupted.
Press 'y' to continue: y
[Notes]
A system restart or shutdown is required to complete the operation.
Device Reference = /dev/nvme0
Results for format: Operation succeeded. See Notes.
[root@htcel02 soft]# reboot
重启系统后,数据库集群root.sh脚本执行完成。
来自 “ ITPUB博客 ” ,链接:http://blog.itpub.net/24585765/viewspace-2374949/,如需转载,请注明出处,否则将追究法律责任。
转载于:http://blog.itpub.net/24585765/viewspace-2374949/