AeroLab supports installing Vector (Proximus) client to Aerospike.
AeroLab supports vector version 0.3.1 by default. Version can be overriden using the --version=... parameter. Note that due to the current state of Vector development, only the default version has been tested to work with AeroLab.
Vector has specific 2-or-more-namespaces requirement from Aerospike servers. An example can be generated as follows:
aerolab conf generate
Tick the vector checkbox and optionally the on-disk checkbox for the vector namespace. Press CTRL+X to save as aerospike.conf.
In this example, create 2 nodes, specifying GCP details. Use the generated aerospike.conf.
aerolab cluster create -n vectordb -c 2 -o aerospike.conf --zone us-central1-a --instance e2-standard-4
aerolab client create vector -n vector -C vectordb --confirm --zone us-central1-a --instance e2-standard-4
The following vector-specific command-line parameters apply to your proximus cluster:
-C, --cluster-name= aerospike cluster name to seed from (default: mydc)
--seed= specify an aerospike cluster seed IP:PORT instead of providing a ClusterName; if this parameter is provided, ClusterName is ignored
--listen= specify a listen IP:PORT for the service (default: 0.0.0.0:5555)
--no-touch-listen set this to prevent aerolab from touching the service: configuration part
--no-touch-seed set this to prevent aerolab from configuring the aerospike seed ip and port
--no-touch-advertised set this to prevent aerolab from configuring the advertised listeners
--version= vector version to install; only 0.3.1 is officially supported by aerolab (0.3.1-1 for rpm) (default: 0.3.1)
--custom-conf= provide a custom aerospike-proximus.yml to ship
--no-start if set, service will not be started after installation
-f, --featurefile= Features file to install; if not provided, the features.conf from the seed aerospike cluster will be taken
--metans= configure the metadata namespace name (default: proximus-meta)
--confirm set this parameter to confirm any warning questions without being asked to press ENTER to continue
The vector client is best paired with a set of examples you can utilize. Follow these instructions to install in the /opt/ director and be invoked with the following steps.
aerolab attach client -n vector
apt -y install python3 python3-pip git
cd /opt && git clone https://github.com/aerospike/proximus-examples.git
Or develop your own application using the aerospike vector python client.
Install the prism example image search application:
aerolab client attach -n vector -- /bin/bash /opt/prism-example.sh install
Once the install finishes, upload images:
aerolab files upload -c -n vector {path-to-pictures-file-or-directory} /opt/proximus-examples/prism-image-search/prism/static/images/data/
Run the web server:
aerolab client attach -n vector -- /bin/bash /opt/prism-example.sh
On docker, access http://127.0.0.1:8998.
On cloud deployments, run aerolab client list and then access http://EXTERNAL_IP:8080.
$ aerolab client attach -n vector
echo "nohup /bin/bash /opt/prism-example.sh >> /var/log/prism.log 2>&1 &" > /opt/autoload/15-prism
chmod 755 /opt/autoload/15-prism
exit
$ aerolab client attach -n vector --detach -- /bin/bash /opt/autoload/15-prism
As the script is in /opt/autoload, prism will also be auto-started whenever aerolab starts the client machine.
The below example is for Docker. For GCP/AWS add the appropriate --zone=, --instance=, --instance-type= to all the cluster/client create commands. See aerolab cluster create help and aerolab client create help for more details.
# make a working directory
mkdir vector
cd vector
# create aerospike.conf - tick the vector and vectod-disk checkboxes
aerolab conf generate
# create aerospike cluster
aerolab cluster create -n vectordb -o aerospike.conf
# add exporter to aerospike cluster for monitoring
aerolab cluster add exporter -n vectordb
# create vector client and configure it to use the aerospike cluster
aerolab client create vector -n vector -C vectordb --confirm
# create AMS monitoring stack, configure to monitor cluster and client, exposing prometheus port in docker
aerolab client create ams -n ams --clusters=vectordb --vector=vector -e 9090:9090
# install prism example in the vector client
aerolab client attach -n vector -- /bin/bash /opt/prism-example.sh install
# upload example images
aerolab files upload -c -n vector {path-to-image-folder-or-file} /opt/proximus-examples/prism-image-search/prism/static/images/data/
# create and upload prism example startup script
echo "nohup /bin/bash /opt/prism-example.sh >> /var/log/prism.log 2>&1 &" > 15-prism
aerolab files upload -c -n vector 15-prism /opt/autoload/15-prism
aerolab attach client -n vector -- chmod 755 /opt/autoload/15-prism
# run the prism example
aerolab client attach -n vector --detach -- /bin/bash /opt/autoload/15-prism
# tail prism logs
aerolab attach client -n vector -- tail -f /var/log/prism.log
On docker, access http://127.0.0.1:8998.
On cloud deployments, run aerolab client list and then access http://EXTERNAL_IP:8080.
For accessing the AMS stack, see “Access URL” output of aerolab client list.
aerolab cluster destroy -f -n vectordb
aerolab client destroy -f -n ams,vector