Spotlight on Oracle 10.10 - Getting Started Guide

The loss or corruption of blocks between instances in a cluster indicates that there may be a problem with the cluster interconnect, the primary means of transporting data between instances in an Oracle RAC environment. The charts on this page –Blocks Lost, Blocks Corrupt, Workload, and Latency – may help indicate the source of any problems with the cluster interconnect.

To open the Corrupt / Lost Blocks page

Select the Spotlight on Oracle RAC connection in the Spotlight Browser.
Click Cluster | Corrupt/Lost Blocks.

Charts on the Corrupt/Lost Blocks page

Notes:

Every chart has a legend (list of symbols) to its right that describes the various series (line graphs) on the chart.
Click an item in the legend to highlight its series (line) in the chart. Click a second time to return the series to its normal appearance.
Move the mouse pointer over an item in the legend to view the current value for that series within the chart.

Chart	Description
Blocks Lost and Blocks Corrupt	These two charts show the current status of requested block transfers across the cluster interconnect. When there is a problem with the loss or corruption of transferred blocks, you should: Verify that a dedicated private interconnect is configured between the nodes participating in the cluster. Verify that you are using appropriate settings for operating system-specific IPC (interprocess communication) buffer sizes for the interconnects. Check for hot blocks being shared between instances. These may be candidates for partitioning. Reduce the number of full table scans that cause large cache movement between instances. Verify scheduling delays and process queue lengths, making changes as required.
Workload	Data is exchanged between Oracle RAC instances when the required blocks are available in the buffer caches of ANY of the participating instances. Only when this data is NOT available in cache does Oracle need to read data from disk. Workload displays block send rate – the amount of data sent by the instances in response to requests received from other instances in the cluster. This consolidated chart shows the workload for each instance, and allows you to compare the details in the Blocks Lost and Blocks Corrupt charts against the corresponding block send rate.
Latency	In an Oracle RAC environment, when a block of data is not found in the local cache of an instance, Oracle first checks the cache of the remaining instances in the cluster before reading the block from disk. The cluster latency is the time taken to receive a block from another instance in the cluster. This consolidated chart shows the recent cluster latency for each instance, and allows you to compare the details in the Blocks Lost and Blocks Corrupt charts against it. Notes: This chart is the same as the Cluster \| Latency Page \| Cluster Latency chart. The performance of the cluster interconnect is measured by monitoring the GCS (Global Cache Service) activity between instances. While high latencies may occur during infrequent spikes in user activity, the average latency should be maintained at a reasonable level. To obtain the best performance, consider using a dedicated private network connection between instances. Depending on the number of instances participating in the cluster, configure this private network either as a direct connection between nodes or via a switch when configured for three or more instances.

Chart

Description

Blocks Lost and Blocks Corrupt

These two charts show the current status of requested block transfers across the cluster interconnect. When there is a problem with the loss or corruption of transferred blocks, you should:

Verify that a dedicated private interconnect is configured between the nodes participating in the cluster.
Verify that you are using appropriate settings for operating system-specific IPC (interprocess communication) buffer sizes for the interconnects.
Check for hot blocks being shared between instances. These may be candidates for partitioning.
Reduce the number of full table scans that cause large cache movement between instances.
Verify scheduling delays and process queue lengths, making changes as required.

Workload

Data is exchanged between Oracle RAC instances when the required blocks are available in the buffer caches of ANY of the participating instances. Only when this data is NOT available in cache does Oracle need to read data from disk.

Workload displays block send rate – the amount of data sent by the instances in response to requests received from other instances in the cluster. This consolidated chart shows the workload for each instance, and allows you to compare the details in the Blocks Lost and Blocks Corrupt charts against the corresponding block send rate.

Latency

In an Oracle RAC environment, when a block of data is not found in the local cache of an instance, Oracle first checks the cache of the remaining instances in the cluster before reading the block from disk.

The cluster latency is the time taken to receive a block from another instance in the cluster. This consolidated chart shows the recent cluster latency for each instance, and allows you to compare the details in the Blocks Lost and Blocks Corrupt charts against it.

Notes:

This chart is the same as the Cluster | Latency Page | Cluster Latency chart.
The performance of the cluster interconnect is measured by monitoring the GCS (Global Cache Service) activity between instances. While high latencies may occur during infrequent spikes in user activity, the average latency should be maintained at a reasonable level.

To obtain the best performance, consider using a dedicated private network connection between instances. Depending on the number of instances participating in the cluster, configure this private network either as a direct connection between nodes or via a switch when configured for three or more instances.

Tip: For complete descriptions of Oracle RAC terms and processes, see the Glossary on the Oracle Real Application Clusters Concepts Web site.

製品を選択してください:

サービス向上のために、「Purpose of your Chat（チャットの目的）」を記入してください。

お客様の不具合に推奨されるソリューション

Spotlight on Oracle 10.10 - Getting Started Guide

Corrupt/Lost Blocks Page