Installing

A Stroom installation is a handful of pieces that have to agree with each other: a database, one or more Stroom nodes, somewhere to put the data, and a way for people to sign in.

Getting them up in the right order saves time, because each one is hard to test until the one before it works.

The order

1. A database. One database serves the whole cluster, not one per node. Build it first, because a Stroom node that cannot reach it will not start.
2. Java and the application on each node. See Requirements.
3. The configuration file on each node, which is where the database connection and anything node-specific lives. See Configuration.
4. The first administrator, from the command line. A new installation has nobody who can sign in, so until this is done the user interface is of no use to you. See The first administrator.
5. Authentication, if you are using an external provider. See Security setup.
6. Volumes, so there is somewhere to write data and index shards. See Volumes.
7. Turn processing on. Pipeline processing ships disabled, so nothing is processed until both the Processor Task Creator and Data Processor jobs are enabled. The first creates the work and the second does it, on each node where it is on. See Scheduled jobs.

Steps 4 and 7 are the two that catch people, because neither announces itself. A system with no administrator looks like a broken login, and a system with processing off looks like a broken pipeline.

Then check it

Post-install checks walks through confirming the installation does what you think, in an order that isolates each piece.

Worth doing even when everything appears fine: most of what Stroom does happens in the background, so "it started without errors" and "it works" are different claims.

Where this fits

This section covers what Stroom needs and the steps that have to happen outside the user interface. How you deliver it, whether that is a container stack, a package on a virtual machine, or an orchestrated cluster, depends on your own estate and standards.

Running a proxy covers standing up a Stroom Proxy, which needs a JVM and a disk and no database at all. Air-gapped installation covers the part that differs on an isolated network, and Hosting your own content store covers mirroring the content packs into a Git server you control. Upgrading covers what to change before moving an existing installation to 7.14, and why the whole cluster has to stop.