The Cluster Realtime Status view contains two categories of information: Host Metrics and Segment Status.
The Host Metrics tab displays a table of the hosts in the cluster with statistics collected at the most recent quantum interval.
At the top right, Last Sync displays the time that the statistics were last updated.
Click a column header to sort the table by that column. Click again to toggle between ascending and descending sort. Coordinator and standby hosts are not included in the sort and are always displayed following the sorted list of segment hosts.
For each server, the following columns are displayed:
top
, there are slight differences.
top
, adjust the
top
CPU user time to be: CPU user time +
nice
time. To compare CPU sys time, before Command Center 4.9/6.1, adjust the
top
utility CPU sys time to be: sys time + iowait time. After 4.10/6.2, the values are similar, for example CPU sys = sys and CPU IOWait = iowait.
Memory is calculated as follows:
Total = MemTotal
Free = MemFree + Buffers + Cached
Used = Total - Free
Committed_AS represents the current amount of memory allocated on the system. It encompasses all memory allocated by processes, regardless of whether it has been utilized by them yet or not.
Disk and Network skew values are calculated by subtracting the median skew value from the current skew value and dividing the result by the median skew value; this is calculated from all segment hosts.
No/Low
Skew value is within +/- 25% of the median.
Moderate
Skew value is between +/- 25% and +/- 50% of the median.
High
Skew value is over +/- 50% of the median..
The Segment Status tab provides a health overview for the VMware Greenplum segments and details for each primary and mirror segment.
VMware Greenplum is most efficient when all segments are operating in their preferred roles. The Segment Summary panel tells you the overall segment status and if any mirrors are acting as primaries.
The Segment Summary panel provides the following information:
The database state can be one of the following:
The number of mirror segments acting as primary segments.
Suggests actions to perform to restore the cluster to balance. These include:
These actions are executed from the command line using the gprecoverseg
Greenplum management utility. See gprecoverseg
in the VMware Greenplum utilities documentation for more information.
The total number of primary and mirror segments in the Greenplum cluster.
The total number of segment hosts in the Greenplum cluster.
The Segment Health panel contains charts for VMware Greenplum segments’ status, replication mode, and preferred roles.
Numbers of segments that are down and up.
A chart that shows the number of segments in each of the possible replication modes.
Not Syncing: The primary segment and mirror segment are active and all changes to the primary segment have been copied to the mirror using a file block replication process.
Change Tracking: If a primary segment is unable to copy changes to its mirror segment using the file replication process, it logs the unsent changes locally so they can be replicated when the mirror again becomes available. This can happen if a mirror segment goes down or if a primary segment goes down and its mirror segment automatically assumes the primary role.
Resyncing: When a down segment is brought back up, administrators initiate a recovery process to return it to operation. The recovery process synchronizes the segment with the active primary and copies the changes missed while the segment was down.
Synced: Once all mirrors and their primaries are synchronized, the system state becomes synchronized.
The red portion of the Preferred Role chart shows the numbers of segments that not operating in their preferred primary or mirror roles. If the chart is not solid green, the performance of the Greenplum cluster is not optimal.
Primary and mirror segments are distributed evenly among the segment hosts to ensure that each host performs an equivalent share of the work and primary segments and their mirror segments reside on different segment hosts. When a primary segment goes down, its mirror on another host in the cluster automatically assumes the primary role, increasing the number of primary segments running on that host. This uneven distribution of the workload will affect query performance until the down segment is restored and the segments are returned to their original, preferred, roles.
The table at the bottom of the Segment Status page contains a detailed row for every primary and mirror segment in the Greenplum Cluster. The table has the following columns for each segment: