Highlighted
8 Krypton

通过vplex分盘,为何有些系统得从lun1开始分

通过vplex分盘,为何有些系统得从lun1开始分,vplex默认是从lun0开始,为何有些系统不认,得从lun1开始呢?

因为最近vplex要升微码,检查后发现有几个view是从lun1开始分的,CE说需要再给这几个view分块盘把lun0占住。请问,

1、如果我给系统分了lun0后,系统只要不去扫盘,这肯定是认不到的,若有系统重启,那么这块lun0应该是能扫到吧,重启后原来powerpath给各个卷的伪名顺序会变吗,新加的这块盘系统会怎么识别,是识不到这块盘呢还是按powerpath命名规则再顺序命名。

2、若现在要求各系统管理员输一下powermt save,至少能保证重启后原来的伪名不变吧?

3、现在想升完微码后再把这些个lun0 remove掉,在系统无感知的条件下,有啥建议吗?

标签 (1)
标记 (1)
0 项奖励
3 条回复3
8 Krypton

Re: 通过vplex分盘,为何有些系统得从lun1开始分

VPLEX LUN0的问题产生的原因和解决方法可以参考这个KB,Why does my host see a LUNZ when it is zoned and logs into the front end port of a VPLEX? https://support.emc.com/kb/92642

产生的原因一般是下面几种:

  • If the initiator has not been registered a LUNZ will be seen.

  • If there is no HLU 0, a LUNZ will be seen.

  • If there is no storage allocated in the storage view a LUNZ will be seen.

  • If the proper FE ports are not in the storage-view a LUNZ will be seen.

解决方法如下:

  1. Register the host(s) initiators on the VPLEX.

  2. Assign storage to the hosts initiator in the VPLEX storage-view

  3. Ensure it has at least 1 LUN with a LUN0 (HLU). The first LUN allocated to the storage view will be allocated LUN0.

  4. Ensure the proper zoned FE ports are in the the correct storage-view.

CE建议你把LUN0占住是因为有一条ETA (EMC Technical Advisories) KB:ETA 190162: VPLEX: Provisioning a LUN ID 0 to all storage views is necessary to address the potential of a Total Cluster Outage (TCO) and data unavailability during a Non-Disruptive Upgrade (NDU). https://support.emc.com/kb/190162 . 主要是为了防止在NDU维护时启用I/O forwarding后可能的数据不可用问题。这一条KB可能会涉及一些列更复杂的操作,建议让CE或EMC工程师来帮忙确认下。

8 Krypton

Re: 通过vplex分盘,为何有些系统得从lun1开始分

俩链接都被拒绝了,大概意思已经知道了。

1111111.png

0 项奖励
8 Krypton

Re: 通过vplex分盘,为何有些系统得从lun1开始分

你帐号绑定VPLEX产品了吗?清掉浏览器缓存再试试?这两条KB都是普通用户就能看的。

0 项奖励