With two small aix machine built a set of oracle rac database, hereinafter referred to as "small machine a and b
Now b small machine with hardware fault detection disaster prevention cure to change hardware, planned downtime repair six hours

Leadership can delay due to rac environment, business, want to stop b small machine directly, outage replacement hardware devices, and the other a normal operation of a

Consult everybody a great god such risk is too big? Will not cause small machine database b can't normally start?

If I do that, small machine b can also normal to join in the cluster?

CodePudding user response:

Yes, that is to build the advantage of cluster,

The original poster can baidu "increase cluster nodes," "cluster" remove node, there should be a lot of articles;

PS: if not sure, suggest to find related experience of assistance,

CodePudding user response:

1, delete the node b;
2, in the node b hardware;
3, increase the new node b
Fix,

But there is a premise that nodes can resist a RAC of a complete set of original to undertake the highest load

CodePudding user response:

refer to the second floor minsic78 response:

1, delete the node b;
2, in the node b hardware;
3, increase the new node b
Fix

But there is a premise that nodes can resist a RAC undertake the highest load of a complete set of original ~

If the replacement hardware does not involve the operating system reinstall software reshipment network card, HBA card replacement, etc., and then turn it off in it directly

CodePudding user response:

Stopped after a node can carry business load to live? Advice when idle,

CodePudding user response:

reference minsic78 reply: 3/f

Quote: refer to the second floor minsic78 response:

1, delete the node b;
2, in the node b hardware;
3, increase the new node b
Fix

But there is a premise that nodes can resist a RAC undertake the highest load of a complete set of original ~

If the replacement hardware does not involve the operating system reinstall software reshipment network card, HBA card replacement, etc., and then shut down directly in

Now system has not yet officially launched, single node of the business, does not involve any software reshipment
If not eliminate nodes directly shutdown in risky?

CodePudding user response:

In addition if sure to perform an abridged points, increase the operation of the node, it is important to note that inspection steps that no pass inspection items, don't think that should not affect someone add node (our side just because this affect the storage disk, the head of state is wrong cause accident).

CodePudding user response:

reference 1st floor wmxcn2000 response:

yes, this is the advantage of cluster setup,

The original poster can baidu "increase cluster nodes," "cluster" remove node, there should be a lot of articles;

PS: if there is no hold, suggest to find related experience,

Just checked the, not the simple,
If this single instance only library stopped, such as replacement of the hardware to rise, there will be a problem?

CodePudding user response:

refer to 6th floor liuzhijian2008x response:

if determined to perform other abridged points, increase the operation of the node, it is important to note that no pass inspection steps of inspection items, don't assume that the should not impact (on our side someone add node for that affect the storage disk, the head of state is wrong cause accident),

Plan is currently the only need to replace the hardware which library to stopped, deleted, before the Spring Festival, now increases the risk of node is too big

CodePudding user response:

Or see you in what hardware, normal change the CPU memory what should be no problem, or you may change anyway in a disk in the RAID 1 is fine in theory, but if you want to change for your software installation path where the disk, and is a single disk has not set any RAID, it would definitely have a problem,

CodePudding user response:

refer to the eighth floor u011962997 response:

Quote: refer to the sixth floor liuzhijian2008x response:

If determined to perform other abridged point, increase the operation of the node, it is important to note that inspection steps that no pass inspection items, don't assume that the should not impact (on our side someone add node for that affect the storage disk, the head of state is wrong cause accident),

Plan is currently the only need to replace the hardware which library to stopped, deleted, before the Spring Festival, now increases the risk of node is too big

9/f, said is right, see you in what hardware,

CodePudding user response:

references 9 f minsic78 response:

or see you in what hardware, normal change the CPU memory what should be no problem, or you may change anyway in a disk in the RAID 1 is fine in theory, but if you want to change for your software installation path in the disk, and is a single disk has not set any RAID, it would definitely have a problem,

Memory, do not affect the disk

CodePudding user response:

Change the memory does not affect cluster, need not abridged point, add nodes, shutdown - "in memory -" startup database,