Hi Madhu, thank you for the question!
Your understanding is correct, the out-of-the-box helm charts come with extended visibility, provided by ElasticSearch (and my understanding is that kafka is used solely for the extended visibility functionality).
If you want to deploy over mysql, you can use the instructions provided in https://github.com/temporalio/helm-charts#bring-your-own-mysql . This will configure the system to use mysql
for both default and visibility data stores, but it will still use ElasticSearch for extended visibility features (such as tctl --ns mynamespace wf count
).
You can also skip ElasticSearch (and ES-only extended visibility functionality) altogether. To do that, provide these two helm install
options. to not include ElasticSearch/kafka in your installation:
--set elasticsearch.enabled=false
--set kafka.enabled=false
Here is an example of a complete helm install
command line that does just that, which is based on what we continuously run in our test pipelines (I also just tested this, just in case;):
~/src/temporal-helm-charts $ helm install \
-f values/values.mysql.yaml temporaltest \
--set server.replicaCount=3 \
--set server.config.persistence.default.sql.user=<MYSQLUSER> \
--set server.config.persistence.default.sql.password=<MYSQLPWD> \
--set server.config.persistence.visibility.sql.user=<MYSQLUSER> \
--set server.config.persistence.visibility.sql.password=<MYSQLPWD> \
--set server.config.persistence.default.sql.host=<MYSQLHOST> \
--set server.config.persistence.visibility.sql.host=<MYSQLHOST> \
--set server.config.persistence.default.sql.database=temporal \
--set server.config.persistence.visibility.sql.database=temporal_visibility \
--set elasticsearch.enabled=false \
--set kafka.enabled=false . \
--timeout 15m
Hope this helps!
Thank you,
Mark.
PS I added a github ticket to also clearly document this in the readme. https://github.com/temporalio/helm-charts/issues/57