pcp-pmda-hdb is a Performance Co-Pilot (PCP) Performance Metric Domain Agent (PMDA) for SAP HANA (hdb).
⚠️ WARNING: Alpha Software
This project is in an early stage and should not be used in productive environments.
pcp-pmda-hdb connects to HANA via the hdbcli Python module.
pip3 install hdbcli
dnf install -y pcp python3-pcp pcp-zeroconf
# get pcp-pmda-hdb
git clone <REPO>
cd pcp-pmda-hdb
# copy files
sudo mkdir -p /var/lib/pcp/pmdas/hdb
sudo cp pmdahdb.py /var/lib/pcp/pmdas/hdb/pmdahdb.python
sudo cp Install Remove pmdahdb.conf /var/lib/pcp/pmdas/hdb/
sudo chown -R root /var/lib/pcp/pmdas/hdb/
cd /var/lib/pcp/pmdas/hdb/
# Set connection parameters in pmdahdb.conf, see Configuration section of this document
sudo $EDITOR pmdahdb.conf
sudo chmod 640 pmdahdb.conf
# actual installation with pcp
sudo ./Install
> Updating the Performance Metrics Name Space (PMNS) ...
> Terminate PMDA if already installed ...
> Updating the PMCD control file, and notifying PMCD ...
> Check hdb metrics have appeared ... 188 metrics and 1522 values
# check hdb-pmda is marked as installed
pcp | grep hdb
List available metrics:
pminfo -t hdb
hdb.io.aysnc_requests_count [Number of active asynchronous input and/or output requests on the host]
hdb.io.file_handles_count [Number of allocated file handles on the host]
hdb.cpu.total_time_idle_milliseconds [CPU idle time in milliseconds]
hdb.cpu.total_time_iowait_milliseconds [CPU time spent waiting for I/O in milliseconds]
hdb.cpu.total_time_system_milliseconds [CPU time spent in system mode in milliseconds]
hdb.cpu.total_time_user_milliseconds [CPU time spent in user mode in milliseconds]
hdb.alerts.active_count [Number of currently active alerts reported by the statistics server per rating]
hdb.memory.oom_events.statement_memory_limit_count [Number of out-of-memory (OOM) events since last reset caused by a statement memory limit.]
hdb.memory.oom_events.process_allocation_limit_count [Number of out-of-memory (OOM) events since last reset caused by a process allocation limit.]
[...]
Read all metrics
pminfo -Ff hdb
hdb.io.aysnc_requests_count
inst [0 or "hxehost"] value 10240
hdb.io.file_handles_count
inst [0 or "hxehost"] value 2816
[...]
Read a metric
pminfo -dfmtT hdb.version
hdb.version PMID: 158.0.2 [HANA Version (Revision)]
Data Type: string InDom: PM_INDOM_NULL 0xffffffff
Semantics: discrete Units: none
Help:
HANA Version (Revision)
value "2.00.054.00.1611906357"
Read a metric with multiple instances
pminfo -dfmtT hdb.services.memory.stack_size_bytes
hdb.services.memory.stack_size_bytes PMID: 158.4.3 [Stack size in bytes]
Data Type: 64-bit unsigned int InDom: 158.3 0x27800003
Semantics: instant Units: byte
Help:
Stack size in bytes
inst [0 or "hxehost.39001.nameserver"] value 148111360
inst [1 or "hxehost.39003.indexserver"] value 129761280
inst [2 or "hxehost.39006.webdispatcher"] value 40894464
inst [3 or "hxehost.39010.compileserver"] value 29360128
The official PCP documentation contains the complete documentation on PCP.
RHEL specific information can be found in the Red Hat Enterprise Linux PCP Documentation
If you want to visualise the metrics via Grafana, have a look at this blog post: Visualizing Performance with Grafana.
pmdahdb.conf
contains the configuration of pcp-pmda-hdb.
Curently there is only one hdb
section that contains the SAP HANA connection parameters.
Quoted passwords are supported.
[hdb]
host=hana-host
port=39015
user=SYSTEM
password="Secret"
pcp-pmda-hdb can be used in scale-out scenarios.
All metrics are Host and Port aware.
Host and Port are included in the instance names (<host>.<port>.<instance>
).
Metric | Description |
---|---|
hdb.admission_control.measurement.memory_allocation_percent | Last measured memory size, as a percentage of the global allocation limit |
hdb.admission_control.measurement.memory_size_gigabytes | Last measured memory size in GB |
hdb.admission_control.measurement.timestamp | The time at which the last memory size was measured |
hdb.admission_control.queue_size | Current waiting request queued count |
hdb.admission_control.total_admit_count | Accumulated request admission count |
hdb.admission_control.total_dequeue_count | Accumulated request dequeued count (the executed request count) |
hdb.admission_control.total_enqueue_count | Accumulated request enqueued count |
hdb.admission_control.total_reject_count | Accumulated request rejection count |
hdb.admission_control.total_timeout_count | Accumulated request dequeued count due to timeout (the rejected request count) |
hdb.admission_control.wait_time.avg_microseconds | Average wait time of the request in the queue in microseconds |
hdb.admission_control.wait_time.last_microseconds | Last wait time of the request in the queue in microseconds |
hdb.admission_control.wait_time.max_microseconds | Maximum wait time of the request in the queue in microseconds |
hdb.admission_control.wait_time.min_microseconds | Minimum wait time of the request in the queue in microseconds |
hdb.admission_control.wait_time.sum_microseconds | Total wait time of the request in the queue in microseconds |
hdb.alerts.active_count | Number of currently active alerts reported by the statistics server per rating |
hdb.backup.cancel_pending_count | Number of currently running backups with pending cancellation |
hdb.backup.max_current_runtime_seconds | Maximum runtime of currently running (state=running or cancel pending) backups |
hdb.backup.max_successful_runtime_seconds | Maximum runtime of successful backups |
hdb.backup.running_count | Number of currently running backups |
hdb.backup.total_canceled_count | Total number of canceled backups |
hdb.backup.total_failed_count | Total number of failed backups |
hdb.backup.total_successful_count | Total number of successful backups |
hdb.buffer_cache.allocated_size_bytes | Allocated memory for the buffer cache in bytes |
hdb.buffer_cache.hit_ratio | Ratio of pages found in the buffer cache to pages requested from the buffer cache |
hdb.buffer_cache.max_size_bytes | Maximum buffer cache memory capacity in bytes |
hdb.buffer_cache.reuse_count | Number of times that a buffer is released for reuse by the cache |
hdb.buffer_cache.used_size_bytes | Used memory for the buffer cache in bytes |
hdb.buffer_cache_pool.buffers_reuse_count | Number of buffers released from the LRU list for the pool so that a requested page can be cached |
hdb.buffer_cache_pool.free_buffers_count | Number of free buffers for the pool |
hdb.buffer_cache_pool.growth_percent | Rate, as a percentage, at which the buffer pool can grow |
hdb.buffer_cache_pool.hot_list_buffers_count | Number of buffers in the hot buffer list for the pool |
hdb.buffer_cache_pool.lru_list_buffers_count | Number of buffers in the LRU chain for the pool |
hdb.buffer_cache_pool.out_of_buffer_event_count | Number number of times that an out-of-buffer situation occurred while requesting buffers from the pool. |
hdb.buffer_cache_pool.total_buffers_count | Number of buffers allocated to the pool |
hdb.cache.entries_count | Number of entries in the cache instance |
hdb.cache.hits_count | Number of cache hits for the cache instance |
hdb.cache.inserts_count | Number of insertions into the cache instance |
hdb.cache.invalidations_count | Number of invalidations in the cache instance |
hdb.cache.miss_count | Number of cache misses for the cache instance |
hdb.cache.total_size_bytes | Maximum available memory budget in bytes available for the cache instance |
hdb.cache.used_size_bytes | Memory in bytes used by the cache instance |
hdb.column_unloads.explicit_count | Number of explicit column unloads |
hdb.column_unloads.low_memory_count | Number of column unloads caused by memory management's automatic shrink on out of memory (OOM) |
hdb.column_unloads.merge_count | Number of column unloads due to merges |
hdb.column_unloads.shrink_count | Number of column unloads caused by a manual shrink |
hdb.connections.idle_count | Total number of connections where no statement is being executed |
hdb.connections.queuing_count | Total number of queued connections |
hdb.connections.running_count | Total number of connections where a statement is being executed |
hdb.cpu.total_time_idle_milliseconds | CPU idle time in milliseconds |
hdb.cpu.total_time_iowait_milliseconds | CPU time spent waiting for I/O in milliseconds |
hdb.cpu.total_time_system_milliseconds | CPU time spent in system mode in milliseconds |
hdb.cpu.total_time_user_milliseconds | CPU time spent in user mode in milliseconds |
hdb.instance_id | SAP instance ID |
hdb.instance_number | Instance number |
hdb.io.aysnc_requests_count | Number of active asynchronous input and/or output requests on the host |
hdb.io.file_handles_count | Number of allocated file handles on the host |
hdb.memory.allocated_bytes | Size of the memory pool for all SAP HANA processes in bytes |
hdb.memory.code_size_bytes | Code size, including shared libraries of SAP HANA processes in bytes |
hdb.memory.host_allocation_limit_bytes | Allocation limit for all processes in bytes |
hdb.memory.host_free_bytes | Free physical memory on the host in bytes |
hdb.memory.host_used_bytes | Used physical memory on the host in bytes |
hdb.memory.oom_events.global_allocation_limit_count | Number of out-of-memory (OOM) events since last reset cause by the global allocation limit. |
hdb.memory.oom_events.process_allocation_limit_count | Number of out-of-memory (OOM) events since last reset caused by a process allocation limit. |
hdb.memory.oom_events.statement_memory_limit_count | Number of out-of-memory (OOM) events since last reset caused by a statement memory limit. |
hdb.memory.peak_used_bytes | Peak memory from the memory pool used by SAP HANA processes since the instance started (this is a sample-based value) in bytes |
hdb.memory.shared_size_bytes | Shared memory size of SAP HANA processes in bytes |
hdb.memory.swap_free_bytes | Free swap memory on the host in bytes |
hdb.memory.swap_used_bytes | Used swap memory on the host in bytes |
hdb.memory.used_bytes | Amount of memory from the memory pool that is currently being used by SAP HANA processes in bytes |
hdb.metadata_locks.total_wait_time_microseconds | Accumulated lock wait time (in microseconds) for metadata locks for all available services from database start up until the current time. |
hdb.metadata_locks.total_waits_count | Accumulated lock wait count for metadata locks for all available services from database start up until the current time. |
hdb.mvcc.acquired_lock_count | Number of acquired records locks |
hdb.mvcc.data_versions_count | Number of all MVCC data versions per service |
hdb.mvcc.metadata_versions_count | Number of all MVCC metadata versions per service |
hdb.mvcc.read_write_lag | Difference between minimum closed write transaction ID and what all transactions can see |
hdb.mvcc.snapshot_lag | Difference between global MVCC timestamp and minimal MVCC timestamp which at least one transaction holds |
hdb.mvcc.versions_count | Number of all MVCC versions on the host |
hdb.plan_cache.cached_plans.count | Total number of cached plans in SQL Plan Cache |
hdb.plan_cache.cached_plans.execution_count | Number of total plan executions for cached plans |
hdb.plan_cache.cached_plans.preparation_count | Total number of plan preparations for cached plans |
hdb.plan_cache.cached_plans.total_cursor_duration_microseconds | Total cursor duration for cached plans in microseconds |
hdb.plan_cache.cached_plans.total_execution_time_microseconds | Total execution time for cached plans in microseconds |
hdb.plan_cache.cached_plans.total_preparation_time_microseconds | Total plan preparation duration for cached plans in microseconds |
hdb.plan_cache.capacity_bytes | SQL Plan Cache capacity in bytes |
hdb.plan_cache.evicted_plans.avg_cache_time_microseconds | Average duration in microseconds between plan cache insertion and eviction |
hdb.plan_cache.evicted_plans.count | Number of evicted plans from SQL Plan Cache |
hdb.plan_cache.evicted_plans.execution_count | Number of total plan executions for evicted plans |
hdb.plan_cache.evicted_plans.preparation_count | Number of total plan preparations for evicted plans |
hdb.plan_cache.evicted_plans.size_bytes | Accumulated total size of evicted plans in bytes |
hdb.plan_cache.evicted_plans.total_cursor_duration_microseconds | Total cursor duration in microseconds for evicted plans |
hdb.plan_cache.evicted_plans.total_execution_time_microseconds | Total execution time in microseconds for evicted plans |
hdb.plan_cache.evicted_plans.total_preparation_time_microseconds | Total duration in microseconds for plan preparation for all evicted plans |
hdb.plan_cache.fill_ratio_percent | SQL Plan Cache fill ratio in percent |
hdb.plan_cache.hit_ratio | SQL Plan Cache hit ratio |
hdb.plan_cache.hits_count | Number of hit counts from SQL Plan Cache |
hdb.plan_cache.lookups_count | Nmber of plan lookup counts from SQL Plan Cache |
hdb.plan_cache.size_bytes | SQL Plan Cache size in bytes |
hdb.record_locks.acquired_count | Number of locks that are currently acquired |
hdb.record_locks.memory_allocated_bytes | Allocated memory for record locks in bytes |
hdb.record_locks.memory_used_bytes | Used memory for record locks in bytes |
hdb.record_locks.total_wait_time_microseconds | Accumulated lock wait time (in microseconds) for record locks for all available services from database start up until the current time. |
hdb.record_locks.total_waits_count | Accumulated lock wait count for record locks for all available services from database start up until the current time. |
hdb.schemas.memory.used_bytes | Total used (column store) memory by schema in bytes |
hdb.schemas.total_merge_count | Total number of delta merges on the schema across all tables and partitions |
hdb.schemas.total_read_count | Total number of read accesses on the schema across all tables and partitions |
hdb.schemas.total_write_count | Total number of write accesses on the schema across all tables and partitions |
hdb.services.cpu_percent | CPU usage percentage of the current process |
hdb.services.cpu_time_milliseconds | CPU usage of the current process since the start in milliseconds as if there would be 1 core |
hdb.services.memory.allocated_free_size_bytes | Allocated free memory in bytes |
hdb.services.memory.code_size_bytes | Code size, including shared libraries, in bytes |
hdb.services.memory.compactors_allocated_size_bytes | Part of the memory pool that can potentially (if unpinned) be freed during a memory shortage in bytes |
hdb.services.memory.compactors_freeable_size_bytes | Part of the memory pool that can be freed during a memory shortage in bytes |
hdb.services.memory.effective_allocation_limit_size_bytes | Effective maximum memory pool size, in bytes, considering the pool sizes of other processes |
hdb.services.memory.fragmented_size_bytes | Memory held by SAP HANA's memory management that cannot be easily reused for new memory allocations in bytes |
hdb.services.memory.guaranteed_size_bytes | Minimum guaranteed memory for the process in bytes |
hdb.services.memory.heap_allocated_size_bytes | Heap memory allocated from the memory pool in bytes |
hdb.services.memory.heap_used_size_bytes | Heap memory used from the memory pool in bytes |
hdb.services.memory.heap_used_size_percent | Heap memory used in percent |
hdb.services.memory.logical_size_bytes | Virtual memory size from the operating system perspective in bytes |
hdb.services.memory.physical_size_bytes | Physical memory size from the operating system perspective in bytes |
hdb.services.memory.shared_allocated_size_bytes | Shared memory allocated from the memory pool in bytes |
hdb.services.memory.shared_used_size_percent | Shared memory used in percent |
hdb.services.memory.stack_size_bytes | Stack size in bytes |
hdb.services.memory.used_size_bytes | Memory in use from the memory pool in bytes |
hdb.services.memory.virtual_address_space_total_size_bytes | Total size of the virtual address space in bytes |
hdb.services.memory.virtual_address_space_used_size_bytes | Used size of the virtual address space in bytes |
hdb.services.open_files_count | Number of open files |
hdb.services.sql_executor_threads_active_count | Number of active SQL Executor threads per indexserver |
hdb.services.sql_executor_threads_inactive_count | Number of inactive SQL Executor threads per indexserver |
hdb.services.threads_active_count | Number of active threads |
hdb.services.threads_total_count | Number of total threads |
hdb.statements.compilation_rate | Current statement preparation count per minute |
hdb.statements.execution_count | Total count of all executed statements for data manipulation, data definition, and system control |
hdb.statements.execution_rate | Current statement execution count per minute |
hdb.statements.peak_compilation_rate | Peak statement preparation count per minute |
hdb.statements.peak_execution_rate | peak statement execution count per minute |
hdb.table_locks.total_wait_time_microseconds | Accumulated lock wait time (in microseconds) for table locks for all available services from database start up until the current time. |
hdb.table_locks.total_waits_count | Accumulated lock wait count for table locks for all available services from database start up until the current time. |
hdb.transactions.aborting_count | Current number of aborting transactions |
hdb.transactions.active_count | Current number of active transactions |
hdb.transactions.active_prepare_count | Current number of transactions in active prepare status |
hdb.transactions.blocked_count | Number of currently blocked transactions waiting for locks |
hdb.transactions.blocked_metadata_count | Number of currently blocked transactions waiting for metadata locks |
hdb.transactions.blocked_record_count | Number of currently blocked transactions waiting for record locks |
hdb.transactions.blocked_table_count | Number of currently blocked transactions waiting for table locks |
hdb.transactions.commit_count | Number of transaction commits |
hdb.transactions.commit_rate | Current number of commits per minute |
hdb.transactions.inactive_count | Current number of inactive transactions |
hdb.transactions.partial_aborting_count | Current number of transactions in partial aborting status |
hdb.transactions.peak_commit_rate | Peak number of commits per minute |
hdb.transactions.peak_rollback_rate | Peak rollback of rollbacks per minute |
hdb.transactions.peak_rollbacks_rate | Current number of rollbacks per minute |
hdb.transactions.peak_transaction_rate | Peak transaction count per minute |
hdb.transactions.peak_update_rate | Peak update transaction count per minute |
hdb.transactions.precomitted_count | Current number of precommitted transactions |
hdb.transactions.rollback_count | Number of transaction rollbacks |
hdb.transactions.transaction_rate | Current transaction count per minute |
hdb.transactions.update_count | Number of update transactions |
hdb.transactions.update_rate | Current update transaction count per minute |
hdb.version | HANA Version (Revision) |
hdb.volumes.data.total_size_bytes | Size of data volumes as reported by the file system in bytes |
hdb.volumes.data.used_size_bytes | Size of used and shadow pages in the data volume files in bytes |
hdb.volumes.fill_ratio | Displays the fill ratio of the data volume |
hdb.volumes.io.active_async_reads_count | Number of active asynchronous reads |
hdb.volumes.io.active_async_writes_count | Number of active asynchronous writes |
hdb.volumes.io.async_reads_trigger_ratio | Trigger-ratio of asynchronous reads |
hdb.volumes.io.async_writes_trigger_ratio | Trigger-ratio of asynchronous writes |
hdb.volumes.io.blocked_write_requests_count | Number of blocked write requests |
hdb.volumes.io.max_blocked_write_requests_count | Maximum number of blocked write requests |
hdb.volumes.io.total_appends_count | Total number of appends |
hdb.volumes.io.total_async_reads_count | Total number of triggered asynchronous reads |
hdb.volumes.io.total_async_writes_count | Total number of triggered asynchronous writes |
hdb.volumes.io.total_failed_reads_count | Total number of failed reads |
hdb.volumes.io.total_failed_writes_count | Total number of failed writes |
hdb.volumes.io.total_full_retry_reads_count | Total number of full retry reads |
hdb.volumes.io.total_full_retry_writes_count | Total number of full retry writes |
hdb.volumes.io.total_read_bytes | Total number of read data in bytes |
hdb.volumes.io.total_read_time_microseconds | Total read time in microseconds |
hdb.volumes.io.total_reads_count | Total number of synchronous reads |
hdb.volumes.io.total_short_reads_count | Total number of reads that read fewer bytes than requested |
hdb.volumes.io.total_short_writes_count | Total number of writes that wrote fewer bytes than requested |
hdb.volumes.io.total_time_microseconds | Total I/O time in microseconds |
hdb.volumes.io.total_write_time_microseconds | Total write time in microseconds |
hdb.volumes.io.total_writes_count | Total number of appends |
hdb.volumes.io.total_written_bytes | Total number of written data in bytes |
hdb.volumes.log.total_size_bytes | Size of log volumes as reported by the file system in bytes |
hdb.volumes.total_size_bytes | Total size of the data volume in bytes |
hdb.volumes.used_size_bytes | Used size of the data volume in bytes |
pcp-pmda-hdb currently does not support Multitenant Database Container (MDC) systems. The agent can only target a single database container at the moment (this container can be a tenant database).
pcp-pmda-hdb is not tested with Python2.
Pull requests are welcome. For major changes, please open an issue first to discuss what you would like to change. Please make sure to update tests as appropriate.
dnf install -y pcp python3-pcp pylint python3-pylint shellcheck
pip3 install hdbcli
pip3 install mypy
pip3 install isort
pip3 install flake8
Unit tests (pmdahdb_test.py
) require a running HANA instance (HANA Express is sufficient).
Connection paramneters need to be set via the environment variables HDB_HOST
, HDB_PORT
, HDB_USER
, and HDB_PASSWORD
.
Installation and Removal procedure can be tested with hack/run-integration-tests.py
Integration with the PCP ecosystem can be interactively tested with hack/debug-pmda.sh
.