Dell VNX1 Host Connectivity Guide for Oracle Solaris - Page 155

Planning the Oracle Solaris Cluster Environment, Planning the Oracle Solaris

Page 155 highlights

Solaris and VPLEX connectivity - All cluster nodes also suffering outage: If all the cluster nodes are impacted by the outage, then when each node reboots it will find that it has no reservation key on the storage and the amnesia protection will prevent it from starting to form a cluster. Only when all nodes have rebooted can a cluster be formed. If one node cannot be restarted for some reason, the cluster will remain offline without manual intervention. The consequence is a loss of service (availability problem) but no data corruption. • Mitigating the problems: • The clustered, highly-available, nature of VPLEX means that a total outage is unlikely, especially when storage cluster is geographically distributed. • The use of a VPLEX LUN as the quorum disk for the Solaris Cluster configuration will ensure that storage split-brain resolution will be performed by the VPLEX, and Solaris Cluster resolution will follow that. • The Solaris Cluster quorum system regularly polls for SCSI reservations. Loss of reservations for any reason, for example reset or power loss by the storage, will be detected. This will force each cluster node to panic due to loss of quorum. The storage outage will provoke a complete cluster outage, but data will not be corrupted. The subsequent full cluster reboot after resolution of the storage outage will replace the reservations and the cluster will reform, providing that all nodes reboot. If one or more nodes does not reboot, the Solaris Cluster amnesia protection will prevent the formation of a cluster without operator intervention. • Issues addressed. There are two areas of concern: • When a distributed (campus) configuration is in use, it is essential that any loss of connectivity between sites ("split-brain") is resolved in the same way for both Solaris Cluster and VPLEX. A situation where the host cluster selects site A to survive, while the storage cluster selects site B, will result in a total outage. • A total reset of a complete VPLEX configuration will result in the loss of SCSI-3 reservation keys and registrations. If undetected, this could result in misoperation of the Solaris Cluster software and potential data corruption. • Additional information • VPLEX is supported with Local topology for Local data center clusters, and Metro topology for campus cluster configurations. • When Oracle ASM is in use, the following Oracle and Solaris settings are required: - Set the Oracle ASM instance parameter _asm_hbeatiowait to 200. Refer to Oracle My Oracle Support (MOS) note 1581684.1 for details and also refer to Oracle bug id: 18554251 for background information. - Ensure that NTP time synchronization is setup properly on the cluster so that the setting in step 1 above will have the intended effect. Oracle Solaris Cluster software performs NTP synchronization across nodes by default. Refer to the following documents for details for your particular Oracle Solaris Cluster version: Oracle Solaris Cluster Concepts Guide at Key Concepts for System Administrators and Application Developers > Cluster Time. Oracle Solaris Cluster Software Installation Guide at Planning the Oracle Solaris Cluster Configuration > Planning the Oracle Solaris Cluster Environment > Network Time Protocol (NTP). VPLEX with Oracle Solaris Cluster support 155

  • 1
  • 2
  • 3
  • 4
  • 5
  • 6
  • 7
  • 8
  • 9
  • 10
  • 11
  • 12
  • 13
  • 14
  • 15
  • 16
  • 17
  • 18
  • 19
  • 20
  • 21
  • 22
  • 23
  • 24
  • 25
  • 26
  • 27
  • 28
  • 29
  • 30
  • 31
  • 32
  • 33
  • 34
  • 35
  • 36
  • 37
  • 38
  • 39
  • 40
  • 41
  • 42
  • 43
  • 44
  • 45
  • 46
  • 47
  • 48
  • 49
  • 50
  • 51
  • 52
  • 53
  • 54
  • 55
  • 56
  • 57
  • 58
  • 59
  • 60
  • 61
  • 62
  • 63
  • 64
  • 65
  • 66
  • 67
  • 68
  • 69
  • 70
  • 71
  • 72
  • 73
  • 74
  • 75
  • 76
  • 77
  • 78
  • 79
  • 80
  • 81
  • 82
  • 83
  • 84
  • 85
  • 86
  • 87
  • 88
  • 89
  • 90
  • 91
  • 92
  • 93
  • 94
  • 95
  • 96
  • 97
  • 98
  • 99
  • 100
  • 101
  • 102
  • 103
  • 104
  • 105
  • 106
  • 107
  • 108
  • 109
  • 110
  • 111
  • 112
  • 113
  • 114
  • 115
  • 116
  • 117
  • 118
  • 119
  • 120
  • 121
  • 122
  • 123
  • 124
  • 125
  • 126
  • 127
  • 128
  • 129
  • 130
  • 131
  • 132
  • 133
  • 134
  • 135
  • 136
  • 137
  • 138
  • 139
  • 140
  • 141
  • 142
  • 143
  • 144
  • 145
  • 146
  • 147
  • 148
  • 149
  • 150
  • 151
  • 152
  • 153
  • 154
  • 155
  • 156

VPLEX with Oracle Solaris Cluster support
155
Solaris and VPLEX connectivity
All cluster nodes also suffering outage:
If all the cluster nodes are impacted by the outage, then when each node reboots it
will find that it has no reservation key on the storage and the amnesia protection
will prevent it from starting to form a cluster. Only when all nodes have rebooted
can a cluster be formed. If one node cannot be restarted for some reason, the cluster
will remain offline without manual intervention. The consequence is a loss of
service (availability problem) but no data corruption.
Mitigating the problems:
The clustered, highly-available, nature of VPLEX means that a total outage is unlikely,
especially when storage cluster is geographically distributed.
The use of a VPLEX LUN as the quorum disk for the Solaris Cluster configuration will
ensure that storage split-brain resolution will be performed by the VPLEX, and Solaris
Cluster resolution will follow that.
The Solaris Cluster quorum system regularly polls for SCSI reservations. Loss of
reservations for any reason, for example reset or power loss by the storage, will be
detected. This will force each cluster node to panic due to loss of quorum. The storage
outage will provoke a complete cluster outage, but data will not be corrupted. The
subsequent full cluster reboot after resolution of the storage outage will replace the
reservations and the cluster will reform, providing that all nodes reboot. If one or more
nodes does not reboot, the Solaris Cluster amnesia protection will prevent the
formation of a cluster without operator intervention.
Issues addressed. There are two areas of concern:
When a distributed (campus) configuration is in use, it is essential that any loss of
connectivity between sites (“split-brain”) is resolved in the same way for both Solaris
Cluster and VPLEX. A situation where the host cluster selects site A to survive, while
the storage cluster selects site B, will result in a total outage.
A total reset of a complete VPLEX configuration will result in the loss of SCSI-3
reservation keys and registrations. If undetected, this could result in misoperation of
the Solaris Cluster software and potential data corruption.
Additional information
VPLEX is supported with Local topology for Local data center clusters, and Metro
topology for campus cluster configurations.
When Oracle ASM is in use, the following Oracle and Solaris settings are required:
Set the Oracle ASM instance parameter
_asm_hbeatiowait
to 200. Refer to Oracle
My Oracle Support (MOS) note 1581684.1 for details and also refer to Oracle bug
id: 18554251 for background information.
Ensure that NTP time synchronization is setup properly on the cluster so that the
setting in step 1 above will have the intended effect. Oracle Solaris Cluster software
performs NTP synchronization across nodes by default. Refer to the following
documents for details for your particular Oracle Solaris Cluster version:
Oracle Solaris Cluster Concepts Guide
at
Key Concepts for System
Administrators and Application Developers
>
Cluster Time
.
Oracle Solaris Cluster Software Installation Guide
at
Planning the Oracle Solaris
Cluster Configuration
>
Planning the Oracle Solaris Cluster Environment
>
Network Time Protocol (NTP)
.