Greenplum Command Center uses the gpcc.conf configuration file to save configuration information entered in the Command Center user interface.
You should not normally edit the gpcc.conf file directly. Instead, modify configuration information in the Command Center user interface. The threshold properties for the Recommendations page reports are an exception and must be added manually. If you edit the file, you must restart Command Center for the new values to take effect. The gpcc.conf file is created in the $HOME/gpmetrics/ directory on the VMware Greenplum coordinator or standby host where you start Command Center. The file is an INI-format configuration file, containing properties defined as key = value entries, one property per line. Properties are organized within sections, with the exception of a handful that have no associated section.
| Property | Description | Section |
|---|---|---|
| min_query_time | Queries shorter than number of seconds specified will not be captured in Query History. | [history] |
| wlm_short_query_threshold | Specifies the minimum number of seconds that a query must run before Command Center applies the action of a matching workload management rule. Use this parameter to prevent Command Center from applying rule actions to short-running queries. The default is 45 seconds. | [wlm] |
| wlm_query_cooldown_time | Specifies the wait time, in seconds, for retrying a failed workload management rule (canceling a query or moving a query to another resource group). Command Center makes 2 retry attempts for failed rules, after waiting the configured amount of time. The default is 15 seconds. | [wlm] |
| allow_anonymous | If true, Command Center users can access the Query Monitor view without logging into Command Center. You can change this setting on the Command Center Admin> Permissions page. |
N/A |
| logon_banner | When present in the gpmetrics configuration file, Command Center displays a login banner with whatever text is passed in as a string. You can use this to, among other things, display a disclaimer stating that only authorized users are permitted access and explaining the penalties of unauthorized access. | N/A |
| resource_queue_import_status | Command Center uses this property to determine whether to offer to import VMware Greenplum resource queues to resource groups when you access the Admin> Workload Mgmt view. The default is false. |
N/A |
| emailFrom | The email address to set on the "From:" line of alert emails. The default is [email protected]. Note: Set the email and smtp properties on the Command Center Admin> Alerts page. |
[alert] |
| emailTo | A comma-delimited list of email addresses to send alert emails. | [alert] |
| smtpUsername | The account name to use when authenticating with the SMTP server. | [alert] |
| smtpServer | The address and port of the SMTP server to use for alert emails. | [alert] |
| smtpPassword | The password used to authenticate the SMTP user with the SMTP server, base 64-encoded. | [alert] |
| age_ratio | Specifies the age a table must reach to prompt Command Center to issue a VACUUM FREEZE recommendation in the Age Rank report on the Recommendations page. You must specify a value between 0 and 1. The default is 0.7. The age is calculated using two VMware Greenplum GUCs: xid_warn_limit and xid_stop_limit. For example, 0.7 means a table is regarded as old enough and in need of VACUUM FREEZE when its age is above 70% * (2^31-1 - xid_warn_limit - xid_stop_limit). |
[recommendations] |
| recommendations_visited | Indicates whether the Recommendations page has been visited. | N/A |
| bloat_threshold | Specifies the minimum percentage of bloat (space occupied by dead tuples) required to include a table with a VACUUM recommendation in the Bloat Rank report on the Recommendations page. The default is 0.1 (10%). |
[recommendations] |
| bloat_vacuum_full_threshold | Specifies the minimum percentage of bloat (combined dead tuples and unused space) required to include a table with a VACUUM FULL recommendation in the Bloat Rank report on the Recommendations page. The default is 0.5 (50%). |
[recommendations] |
| accuracy_threshold | Specifies the minimum inaccuracy percentage required to include a table in the Accuracy% Rank table on the Recommendations page. The default is 0.1 (10%). |
[recommendations] |
| skew_threshold | Specifies the minimum skew percentage required to include a table in the Skew% Rank report on the Recommendations page. The default is 0.1 (10%). |
[recommendations] |
| skew_tuple_per_segment | Specifies a minimum number of tuples per segment to include a table in the Skew% Rank report on the Recommendations page. This parameter helps to avoid reporting tables that have high skew due to small numbers of tuples. The default is 128, which means a table with fewer tuples than 128 times the number of segments will be excluded from the report. |
[recommendations] |
| index_bloat_threshold | Specifies the minimum percentage of index bloat (amount of resusable space in an index) required to include a table with a REINDEX recommendation in the Index Bloat Rank report on the Recommendations page. The default is 0.7 (70%). |
[recommendations] |
| stats_check_interval | Specifies how often -- in seconds -- to refresh the statistics in the Command Center Table Browser. The default is 300. New tables and changed values such as the file size and the last access time may not be seen until stats_check_interval seconds have elapsed. To disable table statistics collection, set this property to -1. |
[table_browser] |
| disable_size_check | When set to true, the table browser will not collect table size information, in order to achieve performance gains when the number of tables is very large. The default is false. |
[table browser] |
| gpssServer | Specifies the hostname or IP address of the GPSS server listening for client requests from Commmand Center. | [gpss] |
| gpssPort | Specifies the port number on which the GPSS server listens for client requests from Command Center. The default is 5000. | [gpss] |
| gpssUsername | Specifies the username for authenticating to the GPSS server. | [gpss] |
| gpssAuthentication | Specifies the authentication method for authenticating to the GPSS server. Accepted values are without-tls, tls, and tls-skip-ca-verify. |
[gpss] |
| gpssCaFile | Specifies the file system path to the Certificate Authority file. This file should be accessible by Command Center on the host where it runs. This entry can be empty if the CA is already in the system pool on that host. | [gpss] |
| enable | Specifies whether Command Center automatically collects performance data and saves it in the $GPCC_HOME/logs/pprof file. The default is true. |
[pprof] |
| interval | Specifies how long Command Center will collect data. It takes a value of datatype time duration; a time duration string is a possibly signed sequence of decimal numbers, each with optional fraction and a unit suffix, such as "30ms", "1.5h" or "2h45m". Valid time units are "s", "m", "h". The default is 1h. | [pprof] |
| cpu_interval | Specifies how long Command Center will conduct CPU profiling. It takes a value of datatype time duration; a time duration string is a possibly signed sequence of decimal numbers, each with optional fraction and a unit suffix, such as "30ms", "1.5h" or "2h45m". Valid time units are "s", "m", "h". The default is 10s. | [pprof] |
| ttl | Specifies time-to-live for profile logs. It takes a value of datatype time duration; a time duration string is a possibly signed sequence of decimal numbers, each with optional fraction and a unit suffix, such as "30ms", "1.5h" or "2h45m". Valid time units are "s", "m", "h". The default is 24h | [pprof] |
| profiles | Specifies a comma-delimited list of benchmarking profile names. Valid values are heap -- which means Command Center collects heap benchmark data, cpu -- which means Command Center collects CPU benchmark data, and heap,cpu -- which means Command Center collects both. The default is heap. |
[pprof] |
| sampling_interval | Specifies, in seconds, how often Command Center agents sample queries. The default is 5. |
[sampling] |
| sampling_threads | Limits how many threads Command Center uses in a one-time sampling. You should only adjust this value if there are severe performance issues caused by the agents on the segment hosts. If the number of query processes exceeds this value, some VMware Greenplum sessions may skip this sampling cycle. Setting this value too high may result in performance degradation. This parameter only takes effect when sampling_mode is set to lite. The default is -1, which means there is no limit. |
[sampling] |
| agent_virt_mem_percent | Specifies the maximum percentage of total virtual memory an agent process can consume. When this limit is reached, Command Center terminates the agent to release the memory and then restarts the agent after approximately 15 seconds. Valid values range from 1 to 99. The default is 50, which is 50 percent of total virtual memory. To disable this feature, set this parameter to -1, or set Memory Limit to Unlimited in the Admin > GPCC Settings page. NOTE: This parameter does not affect the agent on the coordinator host. |
[memory limit] |
| coordinator_agent_reset_time | If you observe that the agent process on the coordinator host is consuming too many resources we recommend that you use this parameter to schedule a restart time. When set, specifies the time at which Command Center should restart the coordinator agent. Permitted values are a value in the range 00:00 - 23:59 or a value in the range 00:00:00 - 23:59:59. NOTE: When the agent on the coordinator is restarted Command Center may lose some metrics data. |
N/A |
| enable_instrument_filter | When enabled -- the default -- the Command Center backend only processes instrument data when a user is viewing from the Query Monitor the query associated with that data. If you change the value of this parameter you must restart Command Center for the new value to take effect. We recommend that you not disable this parameter if the workload contains a lot of short-running queries, as this will cause performance issues. NOTE: Enabling this parameter improves performance. However, when enabled, the "Actual Rows" listed in the visual plan tree for finished plan nodes -- nodes that have already finished by the time the user opens the Query Details page and starts viewing the visual plan -- might not be accurate. | N/A |
