[Other doc versions]
[Doc downloads]
Atlassian Stash is the Git code management solution for enterprise teams. It allows everyone in your organisation to easily collaborate on your Git repositories, while providing enterprise-grade support for:
Atlassian offers two deployment options for Stash, to provide enterprise scaling and infrastructure flexibility, and to give administrators control over how Stash fits into their environment:
For most organizations, a single instance of Stash Server provides good performance. Continue reading this page for guidance on best practices in setting up a Stash server in a production environment.
For larger enterprises that require HA and greater performance at scale, Bitbucket Data Center resources uses a cluster of Stash nodes to provide Active/Active failover, and is the deployment option of choice.
Your single instance of Stash Server can be easily upgraded to Stash Data Center when the time comes.
On this page:
Although Stash can be run on Windows, Linux and Mac systems, for enterprise use we only recommend, and support, Linux. This recommendation is based on our own testing and experience with using Stash.
See the Supported platforms page for details of the supported versions of Java, external databases, web browsers and Git.
See Installing Stash Data Center for detailed information about Stash Data Center requirements.
In general, Stash is very stable and has low memory consumption. There are no scalability limits other than for Git hosting operations (clone in particular). We know this is the scalability limit of the product; the limit is proportional to the number of cores on the system.
As an example, data collected from an internal Stash instance indicate that for a team of approximately 50 developers, with associated continuous integration infrastructure, we see a peak concurrency of 30 simultaneous clone operations and a mean of 2 simultaneous clone operations. We conservatively expect that a customer with similar usage patterns would be capable of supporting 1000 users on a machine with 40 cores and a supporting amount of RAM. While we expect a peak concurrency larger than 40, Stash is designed to queue incoming requests so as to avoid overwhelming the server.
Stash Server – see Stash production server data for data from the Stash production instance we run internally at Atlassian.
Stash Data Center – see Bitbucket Data Center Performance for the results of our performance testing for clusters of different sizes.
If Stash is a critical part of your development workflow, maximizing Stash availability becomes an important consideration.
Stash Server – see High availability for Stash for the background information you need to set up Stash Server in a highly available configuration.
Stash Data Center – see Failover for Bitbucket Data Center for information about how Stash Data Center provides HA and almost instant failover.
Stash is built with enterprise scaling and infrastructure flexibility in mind, giving administrators control over how Stash fits into their environment:
Your single instance of Stash Server can be easily upgraded to Stash Data Center when the time comes.
Stash Server – see Scaling Stash for information about how you can tune your Stash server to grow with your organisation's needs. See also Scaling Stash for Continuous Integration performance for information specific to Stash performance when CI tools poll Stash for changes.
Stash Data Center – see Adding cluster nodes to Stash Data Center for information about how you can rapidly provision extra capacity without downtime.
Some possible approaches to provisioning Stash include:
When setting up Stash for a production or enterprise environment, we highly recommend that you configure the following aspects:
<Stash home directory>
/log
. Logs for the bundled Tomcat webserver can be found in <Stash installation directory>
/log
. See Stash debug logging.<Stash home directory
>/audit/logs
directory. Note that Stash has an upper limit to the number of log files it maintains, and deletes the oldest file when a new file is created – we recommend an automated backup of log files. See Audit logging in Stash.