Adding support clickhouse cluster

[preved911]

Hello. I would like to create cluster wide schema_migrations table, when I run migrations on clusterized clickhouse. As I think, It's will guarante that new migrations may be executed on any node of cluster.

[coveralls]

Pull Request Test Coverage Report for Build 1159872900

• 10 of 16 (62.5%) changed or added relevant lines in 1 file are covered.
• No unchanged relevant lines lost coverage.
• Overall coverage decreased (-0.03%) to 57.425%

---

Changes Missing Coverage	Covered Lines	Changed/Added Lines	%
database/clickhouse/clickhouse.go	10	16	62.5%

Totals
Change from base Build 1153229133:	-0.03%
Covered Lines:	3693
Relevant Lines:	6431

---

💛 - Coveralls

[dhui]

For my future reference: clickhouse distributed ddl docs

Overall, this change LGTM. Keep in mind that you may have issues if migrations are run simultaneously on different machines since the clickhouse db driver doesn't use a distributed lock.

Can you add tests or is it not possible to setup a cluster in docker?

[preved911]

Can you add tests or is it not possible to setup a cluster in docker?

@dhui, for me its look like too hard. I viewed realized tests in project and can't imagine how to do that in such flow. For replicate tables over clickhouse nodes, we have to up 2 or more clickhouse nodes with customized config file(need zookeeper instance address specify). dktest, if I undestand correct, do not allow to mount volumes into declared containers. So, should I create very custom test for this case?

[dhui]

Yeah, I don't think it's a trivial amount of work to run multiple containers & images on the same network so we can test a clickhouse cluster setup.

I'd love to have integration tests for clickhouse clusters, but if it's not feasible, we can add a caveat in the docs stating that clickhouse clusters aren't officially supported since there aren't any tests.

[preved911]

@dhui, I added notes in README.md. Did you mean it?

[preved911]

@dhui, so can it be merged?

[dhui]

Sorry for the delay and thanks again for the PR!