Description
Installation
Running
Grafana
Troubleshooting
Operating principles
A Prometheus exporter for Oracle modeled after the MySQL exporter. I'm not a DBA or seasoned Go developer so PRs definitely welcomed.
The following metrics are exposed currently.
- oracledb_exporter_last_scrape_duration_seconds
- oracledb_exporter_last_scrape_error
- oracledb_exporter_scrapes_total
- oracledb_up
- oracledb_activity_execute_count
- oracledb_activity_parse_count_total
- oracledb_activity_user_commits
- oracledb_activity_user_rollbacks
- oracledb_sessions_activity
- oracledb_wait_time_application
- oracledb_wait_time_commit
- oracledb_wait_time_concurrency
- oracledb_wait_time_configuration
- oracledb_wait_time_network
- oracledb_wait_time_other
- oracledb_wait_time_scheduler
- oracledb_wait_time_system_io
- oracledb_wait_time_user_io
- oracledb_tablespace_bytes
- oracledb_tablespace_max_bytes
- oracledb_tablespace_free
- oracledb_tablespace_used_percent
- oracledb_process_count
- oracledb_resource_current_utilization
- oracledb_resource_limit_value
You can run via Docker using an existing image. Since version 0.4, the images are available on the github registry.
Here an example to retrieve the version 0.4.3:
docker pull ghcr.io/iamseth/oracledb_exporter:0.4.3
And here a command to run it and forward the port:
docker run -it --rm -p 9161:9161 ghcr.io/iamseth/oracledb_exporter:0.4.3
If you don't already have an Oracle server, you can run one locally in a container and then link the exporter to it.
docker run -d --name oracle -p 1521:1521 wnameless/oracle-xe-11g-r2:18.04-apex
docker run -d --name oracledb_exporter --link=oracle -p 9161:9161 -e DATA_SOURCE_NAME=system/oracle@oracle/xe ghcr.io/iamseth/oracledb_exporter:0.4.3
Since 0.2.1, the exporter image exist with Alpine flavor. Watch out for their use. It is for the moment a test.
docker run -d --name oracledb_exporter --link=oracle -p 9161:9161 -e DATA_SOURCE_NAME=system/oracle@oracle/xe iamseth/oracledb_exporter:alpine
Different Linux Distros:
x.y.z
- Ubuntu Linux imagex.y.z-oraclelinux
- Oracle Enterprise Linux imagex.y.z-Alpine
- Alpine Linux image
Forked Version:
All the above docker images have a duplicate image tag ending in
_legacy-tablespace
. These versions use the older/deprecated tablespace
utilization calculation based on the aggregate sum of file sizes in a given
tablespace. The newer mechanism takes into account block sizes, extents, and
fragmentation aligning with the same metrics reported from the Oracle Enterprise
Manager. See iamseth#153 for
details. The versions above should have a more useful tablespace utilization
calculation going forward.
Pre-compiled versions for Linux 64 bit and Mac OSX 64 bit can be found under releases.
In order to run, you'll need the Oracle Instant Client Basic for your operating system. Only the basic version is required for execution.
Ensure that the environment variable DATA_SOURCE_NAME is set correctly before starting.
DATA_SOURCE_NAME should be in Oracle EZCONNECT format:
https://docs.oracle.com/en/database/oracle/oracle-database/19/netag/configuring-naming-methods.html#GUID-B0437826-43C1-49EC-A94D-B650B6A4A6EE
19c Oracle Client supports enhanced EZCONNECT, you are able to failover to standby DB or gather some heavy metrics from active standby DB and specify some additional parameters. Within 19c client you are able to connect 12c primary/standby DB too :)
For Example:
# export Oracle location:
export DATA_SOURCE_NAME=system/password@oracle-sid
# or using a complete url:
export DATA_SOURCE_NAME=user/password@//myhost:1521/service
# 19c client for primary/standby configuration
export DATA_SOURCE_NAME=user/password@//primaryhost:1521,standbyhost:1521/service
# 19c client for primary/standby configuration with options
export DATA_SOURCE_NAME=user/password@//primaryhost:1521,standbyhost:1521/service?connect_timeout=5&transport_connect_timeout=3&retry_count=3
# 19c client for ASM instance connection (requires SYSDBA)
export DATA_SOURCE_NAME=user/password@//primaryhost:1521,standbyhost:1521/+ASM?as=sysdba
# Then run the exporter
/path/to/binary/oracledb_exporter --log.level error --web.listen-address 0.0.0.0:9161
Make sure to grant SYS
privilege on SELECT
statement for the monitoring user, on the following tables.
dba_tablespace_usage_metrics
dba_tablespaces
v$system_wait_class
v$asm_diskgroup_stat
v$datafile
v$sysstat
v$process
v$waitclassmetric
v$session
v$resource_limit
Create oracledb_exporter user with disabled login and oracledb_exporter group
mkdir /etc/oracledb_exporter
chown root:oracledb_exporter /etc/oracledb_exporter
chmod 775 /etc/oracledb_exporter
Put config files to /etc/oracledb_exporter
Put binary to /usr/local/bin
Create file /etc/systemd/system/oracledb_exporter.service with the following content:
[Unit]
Description=Service for oracle telemetry client
After=network.target
[Service]
Type=oneshot
#!!! Set your values and uncomment
#User=oracledb_exporter
#Group=oracledb_exporter
#Environment="DATA_SOURCE_NAME=dbsnmp/Bercut01@//primaryhost:1521,standbyhost:1521/myservice?transport_connect_timeout=5&retry_count=3"
#Environment="LD_LIBRARY_PATH=/u01/app/oracle/product/19.0.0/dbhome_1/lib"
#Environment="ORACLE_HOME=/u01/app/oracle/product/19.0.0/dbhome_1"
#Environment="CUSTOM_METRICS=/etc/oracledb_exporter/custom-metrics.toml"
ExecStart=/usr/local/bin/oracledb_exporter \
--default.metrics "/etc/oracledb_exporter/default-metrics.toml" \
--log.level error --web.listen-address 0.0.0.0:9161
[Install]
WantedBy=multi-user.target
Then tell System D to read files:
systemctl daemon-reload
Start this new service:
systemctl start oracledb_exporter
Check service status:
systemctl status oracledb_exporter
Usage of oracledb_exporter:
--log.format value
If set use a syslog logger or JSON logging. Example: logger:syslog?appname=bob&local=7 or logger:stdout?json=true. Defaults to stderr.
--log.level value
Only log messages with the given severity or above. Valid levels: [debug, info, warn, error, fatal].
--custom.metrics string
File that may contain various custom metrics in a TOML file.
--default.metrics string
Default TOML file metrics.
--web.systemd-socket
Use systemd socket activation listeners instead of port listeners (Linux only).
--web.listen-address string
Address to listen on for web interface and telemetry. (default ":9161")
--web.telemetry-path string
Path under which to expose metrics. (default "/metrics")
--database.maxIdleConns string
Number of maximum idle connections in the connection pool. (default "0")
--database.maxOpenConns string
Number of maximum open connections in the connection pool. (default "10")
--web.config.file
Path to configuration file that can enable TLS or authentication.
This exporter comes with a set of default metrics defined in default-metrics.toml. You can modify this file or
provide a different one using default.metrics
option.
NOTE: Do not put a
;
at the end of your SQL queries as this will NOT work.
This exporter does not have the metrics you want? You can provide new one using TOML file. To specify this file to the exporter, you can:
- Use
--custom.metrics
flag followed by the TOML file - Export CUSTOM_METRICS variable environment (
export CUSTOM_METRICS=my-custom-metrics.toml
)
This file must contain the following elements:
- One or several metric section (
[[metric]]
) - For each section a context, a request and a map between a field of your request and a comment.
Here's a simple example:
[[metric]]
context = "test"
request = "SELECT 1 as value_1, 2 as value_2 FROM DUAL"
metricsdesc = { value_1 = "Simple example returning always 1.", value_2 = "Same but returning always 2." }
This file produce the following entries in the exporter:
# HELP oracledb_test_value_1 Simple example returning always 1.
# TYPE oracledb_test_value_1 gauge
oracledb_test_value_1 1
# HELP oracledb_test_value_2 Same but returning always 2.
# TYPE oracledb_test_value_2 gauge
oracledb_test_value_2 2
You can also provide labels using labels field. Here's an example providing two metrics, with and without labels:
[[metric]]
context = "context_no_label"
request = "SELECT 1 as value_1, 2 as value_2 FROM DUAL"
metricsdesc = { value_1 = "Simple example returning always 1.", value_2 = "Same but returning always 2." }
[[metric]]
context = "context_with_labels"
labels = [ "label_1", "label_2" ]
request = "SELECT 1 as value_1, 2 as value_2, 'First label' as label_1, 'Second label' as label_2 FROM DUAL"
metricsdesc = { value_1 = "Simple example returning always 1.", value_2 = "Same but returning always 2." }
This TOML file produce the following result:
# HELP oracledb_context_no_label_value_1 Simple example returning always 1.
# TYPE oracledb_context_no_label_value_1 gauge
oracledb_context_no_label_value_1 1
# HELP oracledb_context_no_label_value_2 Same but returning always 2.
# TYPE oracledb_context_no_label_value_2 gauge
oracledb_context_no_label_value_2 2
# HELP oracledb_context_with_labels_value_1 Simple example returning always 1.
# TYPE oracledb_context_with_labels_value_1 gauge
oracledb_context_with_labels_value_1{label_1="First label",label_2="Second label"} 1
# HELP oracledb_context_with_labels_value_2 Same but returning always 2.
# TYPE oracledb_context_with_labels_value_2 gauge
oracledb_context_with_labels_value_2{label_1="First label",label_2="Second label"} 2
Last, you can set metric type using metricstype field.
[[metric]]
context = "context_with_labels"
labels = [ "label_1", "label_2" ]
request = "SELECT 1 as value_1, 2 as value_2, 'First label' as label_1, 'Second label' as label_2 FROM DUAL"
metricsdesc = { value_1 = "Simple example returning always 1 as counter.", value_2 = "Same but returning always 2 as gauge." }
# Can be counter or gauge (default)
metricstype = { value_1 = "counter" }
This TOML file will produce the following result:
# HELP oracledb_test_value_1 Simple test example returning always 1 as counter.
# TYPE oracledb_test_value_1 counter
oracledb_test_value_1 1
# HELP oracledb_test_value_2 Same test but returning always 2 as gauge.
# TYPE oracledb_test_value_2 gauge
oracledb_test_value_2 2
You can find here a working example of custom metrics for slow queries, big queries and top 100 tables.
If you run the exporter as a docker image and want to customize the metrics, you can use the following example:
FROM iamseth/oracledb_exporter:latest
COPY custom-metrics.toml /
ENTRYPOINT ["/oracledb_exporter", "--custom.metrics", "/custom-metrics.toml"]
NOTE: This has been tested with v0.2.6a and will most probably work on versions above.
NOTE: While
user/password@//database1.example.com:1521,database3.example.com:1521/DBPRIM
works with SQLPlus, it doesn't seem to work with oracledb-exporter v0.2.6a.
In some cases, one might want to scrape metrics from the currently available database when having a active-passive replication setup.
This will try to connect to any available database to scrape for the metrics. With some replication options, the secondary database is not available when replicating. This allows the scraper to automatically fall back in case of the primary one failing.
This example allows to achieve this:
- tns_admin folder:
/path/to/tns_admin
- tnsnames.ora file:
/path/to/tns_admin/tnsnames.ora
Example of a tnsnames.ora file:
database =
(DESCRIPTION =
(ADDRESS_LIST =
(ADDRESS = (PROTOCOL = TCP)(HOST = database1.example.com)(PORT = 1521))
(ADDRESS = (PROTOCOL = TCP)(HOST = database2.example.com)(PORT = 1521))
)
(CONNECT_DATA =
(SERVICE_NAME = DBPRIM)
)
)
TNS_ENTRY
: Name of the entry to use (database
in the example file above)TNS_ADMIN
: Path you choose for the tns admin folder (/path/to/tns_admin
in the example file above)DATA_SOURCE_NAME
: Datasource pointing to theTNS_ENTRY
(user/password@database
in the example file above)
First, set the following variables:
export WALLET_PATH=/wallet/path/to/use
export TNS_ENTRY=tns_entry
export DB_USERNAME=db_username
export TNS_ADMIN=/tns/admin/path/to/use
Create the wallet and set the credential:
mkstore -wrl $WALLET_PATH -create
mkstore -wrl $WALLET_PATH -createCredential $TNS_ENTRY $DB_USERNAME
Then, update sqlnet.ora:
echo "
WALLET_LOCATION = (SOURCE = (METHOD = FILE) (METHOD_DATA = (DIRECTORY = $WALLET_PATH )))
SQLNET.WALLET_OVERRIDE = TRUE
SSL_CLIENT_AUTHENTICATION = FALSE
" >> $TNS_ADMIN/sqlnet.ora
To use the wallet, use the wallet_location parameter. You may need to disable ssl verification with the ssl_server_dn_match parameter.
Here a complete example of string connection:
DATA_SOURCE_NAME=username/password@tcps://dbhost:port/service?ssl_server_dn_match=false&wallet_location=wallet_path
For more details, have a look at the following location: iamseth#84
An example Grafana dashboard is available here.
To build Ubuntu and Alpine image, run the following command:
make docker
You can also build only Ubuntu image:
make ubuntu-image
Or Alpine:
make alpine-image
Retrieve Oracle RPMs (version x.y):
make download-rpms
Then run build:
make linux
Stollen from iamseth#40
First, download Oracle Instant Client 64-Bit version basic and sdk versions.
Extract client (for example: C:\oracle\instantclient_18_5) and extract SDK to the same folder (C:\oracle\instantclient_18_5\sdk)
Set the environment variables:
setx CGO_CFLAGS "C:\oracle\instantclient_18_5\sdk\include"
setx CGO_LDFLAGS "-LC:\oracle\instantclient_18_5 -loci"
Then install GCC (like MSYS2 64 bit in c:\msys64)
Run the MSYS2 MINGW64 terminal and set dependencies packages:
-
Update pacman:
pacman -Su
-
Close terminal and open a new terminal
-
Update all other packages:
pacman -Su
-
Install pkg-config and gcc:
pacman -S mingw64/mingw-w64-x86_64-pkg-config mingw64/mingw-w64-x86_64-gcc
Go to the pkg-config dir c:/msys64/mingw64/lib/pkgconfig/ and create oci8.pc with the following content:
prefix=C:\oracle\instantclient_18_5/sdk/
version=18.5
build=client64
libdir=C:\oracle\instantclient_18_5/sdk/lib/msvc
includedir=C:\oracle\instantclient_18_5/sdk/include
glib_genmarshal=glib-genmarshal
gobject_query=gobject-query
glib_mkenums=glib-mkenums
Name: oci8
Description: Oracle database engine
Version: ${version}
Libs: -L${libdir} -loci
Libs.private:
Cflags: -I${includedir}
Set %PKG_CONFIG_PATH% as the environment variable:
setx PKG_CONFIG_PATH "C:\msys64\mingw64\lib\pkgconfig"
Ensure, that %PATH% includes path to the msys64 binares, if not set it: setx path "%path%;C:\msys64\mingw64\bin"
Everything must compile, including mattn driver for oracle.
Next build ./... in oracledb-exporter dir, or install it.
Oracle is trying to send a value that we cannot convert to float. This could be anything like 'UNLIMITED' or 'UNDEFINED' or 'WHATEVER'.
In this case, you must handle this problem by testing it in the SQL request. Here an example available in default metrics:
[[metric]]
context = "resource"
labels = [ "resource_name" ]
metricsdesc = { current_utilization= "Generic counter metric from v$resource_limit view in Oracle (current value).", limit_value="Generic counter metric from v$resource_limit view in Oracle (UNLIMITED: -1)." }
request="SELECT resource_name,current_utilization,CASE WHEN TRIM(limit_value) LIKE 'UNLIMITED' THEN '-1' ELSE TRIM(limit_value) END as limit_value FROM v$resource_limit"
If the value of limite_value is 'UNLIMITED', the request send back the value -1.
You can increase the log level (--log.level debug
) in order to get the statement generating this error.
error while loading shared libraries: libclntsh.so.xx.x: cannot open shared object file: No such file or directory
This exporter use libs from Oracle in order to connect to Oracle Database. If you are running the binary version, you must install the Oracle binaries somewhere on your machine and you must install the good version number. If the error talk about the version 18.3, you must install 18.3 binary version. If it's 12.2, you must install 12.2.
An alternative is to run this exporter using a Docker container. This way, you don't have to worry about Oracle binaries version as they are embedded in the container.
Here an example to run this exporter (to scrap metrics from system/oracle@//host:1521/service-or-sid) and bind the exporter port (9161) to the global machine:
docker run -it --rm -p 9161:9161 -e DATA_SOURCE_NAME=system/oracle@//host:1521/service-or-sid iamseth/oracledb_exporter:0.2.6a
If you experience an error Error scraping for wait_time: sql: Scan error on column index 1: converting driver.Value type string (",01") to a float64: invalid syntax source="main.go:144"
you may need to set the NLS_LANG variable.
export NLS_LANG=AMERICAN_AMERICA.WE8ISO8859P1
export DATA_SOURCE_NAME=system/oracle@myhost
/path/to/binary --log.level error --web.listen-address :9161
If using Docker, set the same variable using the -e flag.
As being said, Oracle instance may (and probably does) generate a lot of trace files alongside its alert log file, one trace file per scraping event. The trace file contains the following lines
...
*** MODULE NAME:(prometheus_oracle_exporter-amd64@hostname)
...
kgxgncin: clsssinit: CLSS init failed with status 3
kgxgncin: clsssinit: return status 3 (0 SKGXN not av) from CLSS
The root cause is Oracle's reaction of quering ASM-related views without ASM used. The current workaround proposed is to setup a regular task to cleanup these trace files from the filesystem, as example
$ find $ORACLE_BASE/diag/rdbms -name '*.tr[cm]' -mtime +14 -delete
Apache Exporter supports TLS and basic authentication. This enables better control of the various HTTP endpoints.
To use TLS and/or basic authentication, you need to pass a configuration file
using the --web.config
parameter. The format of the file is described
in the exporter-toolkit repository.
Note that the TLS and basic authentication settings affect all HTTP endpoints: /metrics for scraping, /probe for probing, and the web UI.
This exporter supports the multi-target pattern. This allows running a single instance of this exporter for multiple Oracle targets.
To use the multi-target functionality, send a http request to the endpoint /scrape?target=foo:1521
where target is set to the DSN of the Oracle instance to scrape metrics from.