Sonatype Nexus Repository System Requirements
Sonatype fully supports versions of repository manager for one year after the release date. Older releases are supported on a best effort basis and the release dates are listed in our download archives. The terms of support are explained in section 3 of the Support Policy.
Host Operating System
Any Windows, Linux or Macintosh operating system that can run a supported Java version will work. Other operating systems may work, but they are not tested by Sonatype.
The most widely used operating system for Nexus Repository Manager (NXRM) is Linux and therefore customers should consider it the best tested platform.
Dedicated Operating System User Account
Unless you are just testing the repository manager or running it only for personal use, a dedicated operating system user account is strongly recommended to run each unique process on a given host.
The NXRM process user is typically named 'nexus' and must be able to create a valid shell.
As a security precaution, do not run Nexus Repository Manager 3 as the
Adequate File Handle Limits
NXRM3 will most likely want to consume more file handles than the per user default value allowed by your Linux or OSX operating system.
Running out of file descriptors can be disastrous and will most probably lead to data loss. Make sure to increase the limit on the number of open files descriptors for the user running Nexus Repository Manager permanently to 65,536 or higher prior to starting.
See https://issues.sonatype.org/browse/NEXUS-12041 for additional background.
On most Linux systems, persistent limits can be set for a particular user by editing the
/etc/security/limits.conf file. To set the maximum number of open files for both soft and hard limits for the
nexus user to 65536, add the following line to the
/etc/security/limits.conf file, where "nexus" should be replaced with the user ID that is being used to run the repository manager:
nexus - nofile 65536
This change will only take effect the next time the
nexus process user opens a new session. Which essentially means that you will need to restart NXRM.
On Ubuntu systems there is a caveat: Ubuntu ignores the
/etc/security/limits.conf file for processes started by
So if NXRM is started using init.d there, edit
/etc/pam.d/common-session and uncomment the following line ( remove the hash # and space at the beginning of the line):
# session required pam_limits.so
For more information refer to your specific operating system documentation.
If you're using
systemd to launch the server the above won't work. Instead, modify the configuration file to add a
[Unit] Description=nexus service After=network.target [Service] Type=forking LimitNOFILE=65536 ExecStart=/opt/nexus/bin/nexus start ExecStop=/opt/nexus/bin/nexus stop User=nexus Restart=on-abort [Install] WantedBy=multi-user.target
The method to modify the file descriptor limits on OSX has changed a few times over the years. Please note your OS X version and follow the appropriate instructions.
For OS X Yosemite (10.10) and newer
Create the file:
<!DOCTYPE plist PUBLIC "-//Apple//DTD PLIST 1.0//EN" "http://www.apple.com/DTDs/PropertyList-1.0.dtd"> <plist version="1.0"> <dict> <key>Label</key> <string>limit.maxfiles</string> <key>ProgramArguments</key> <array> <string>launchctl</string> <string>limit</string> <string>maxfiles</string> <string>65536</string> <string>65536</string> </array> <key>RunAtLoad</key> <true/> <key>ServiceIPC</key> <false/> </dict> </plist>
If this file already exists, then ensure the value is at least 65536 as shown.
The file must be owned by
root:wheeland have permissions
sudo chmod 644 /Library/LaunchDaemons/limit.maxfiles.plist sudo chown root:wheel /Library/LaunchDaemons/limit.maxfiles.plist
Reboot the operating system to activate the change.
Add a new line to
Restart Nexus Repository to activate the change.
For OS X Lion (10.7) up to OS X Mavericks (10.9)
Create and edit the system file
/etc/launchd.confusing this command:
sudo sh -c 'echo "limit maxfiles 65536 65536" >> /etc/launchd.conf'
Reboot the operating system to activate the change.
Add a new line to
Restart Nexus Repository to activate the change.
Windows operating systems do not need file handle limit adjustments.
The Nexus Repository Docker images are configured with adequate file limits. Some container platforms such as Amazon ECS will override the default limits. On these platforms it is recommended that the Docker image be run with the following flags:
Nexus Repository Manager requires a Java 8 Runtime Environment (JRE). The distributions for OSX and Windows include suitable runtime environments for the specific operating system. The distributions for Unix do not include the runtime environment. If you prefer to use an external runtime or use a Unix operating system, you can choose to install the full JDK or the JRE only. You can confirm the installed Java version with the
java -version command, for example:
$ java -version openjdk version "1.8.0_191" OpenJDK Runtime Environment (build 1.8.0_191-b12) OpenJDK 64-Bit Server VM (build 25.191-b12, mixed mode)
When multiple JDK or JRE versions are installed, you need to ensure the correct version is configured by running the above command as the operating system user that is used to run the repository manager.
In the event you have a non-standard location you need to update the configuration to specify a specific JDK or JRE installation path. To set the path for a specific Java location open the
bin/nexus script and locate the line
INSTALL4J_JAVA_HOME_OVERRIDE. Remove the hash and specify the location of your JDK/JRE:
The startup script verifies the runtime environment by checking for the existence of the nested
bin/java command as well as major and minor version of the runtime to be the required
1.8. If the configured runtime is not suitable, it will proceed with a best effort to locate a suitable runtime configured on the path or via the
JAVA_HOME environment variable. If successful, it will start up the repository manager with this JVM. This allows you to have a dedicated runtime environment for the repository manager installed that is not on the path and not used by other installed applications. Further, you can separate upgrades of the Java runtime used by the repository manager from upgrades of the runtime used by other applications.
Performance is primarily bounded by IO (disk and network) rather than CPU. Available CPUs will impact longer running operations and also the thread allocation algorithms of the web container.
Minimum CPUs: 4
Recommended CPUs: 8+
Configurable Memory Types
Visit the Configuring the Runtime Enviroment page to learn how to change the default memory settings.
JVM Heap Memory
Heap memory stores runtime application objects. A min ( -Xms ) and max ( -Xmx ) value must be specified and the values should be identical.
Increasing the heap memory larger than recommendations or setting the min and max values to be different is not recommended. This will create performance issues causing the operating system to thrash needlessly.
JVM Direct Memory
Only required for OrientDB.
Direct memory is allocated outside of and distinct from heap memory. A max value must be configured if using OrientDB.
Host Physical Memory
The total memory allocated to the entire operating system or virtual hardware, commonly referred to as RAM.
The requirements assume there are no other significant memory hungry processes running on the same host.
|JVM Heap||JVM Direct||Host Physical/RAM|
|Minimum ( default )||2703M||2703M||8G|
|Maximum ( typical )||8G||(host physical/RAM * 2/3) - JVM max heap||no limit|
General Memory Guidelines
- minimum physical/RAM memory on the host 8G
- minimum heap ( -Xms ) must equal set maximum heap ( -Xmx )
- minimum heap size 2703M
- maximum heap size <= 8G
- minimum direct memory ( -XX:MaxDirectMemorySize ) size 2703M
- minimum unallocated host physical/RAM memory should be no less than 1/3 of total physical RAM to allow for virtual memory swap
- max heap + max direct memory <= host physical/RAM * 2/3
Instance Memory Sizing Profiles
These profiles help gauge the typical physical memory requirements needed for a dedicated server host running repository manager. Due to the inherent complexities of use cases, one size does not fit all and this should only be interpreted as a guideline.
|Profile Use Case||Physical/RAM Memory|
repositories < 20
repositories < 50
repositories > 50
Example Maximum Memory Configurations
-XX:MaxDirectMemorySizeis for OrientDB only
Example Maximum Memory Configuration
-Xms2703M -Xmx2703M -XX:MaxDirectMemorySize=2703M
-Xms4G -Xmx4G -XX:MaxDirectMemorySize=4014M
-Xms4G -Xmx4G -XX:MaxDirectMemorySize=6717M
-Xms6G -Xmx6G -XX:MaxDirectMemorySize=15530M
Advanced Database Memory Tuning
Refer to another article which outlines additional memory tuning procedures.
Also see Additional System Requirements for PostgreSQL for further database requirement information, including how to ensure you have the required trigram module for PostgreSQL.
We highly recommend that you have your Nexus Repository 3 instance use an external PostgreSQL database. See our documentation on configuring Nexus Repository Pro for an external PostgreSQL database or on migrating an existing Nexus Repository 3 instance to a PostgreSQL database.
We strongly recommend against running Nexus Repository 3 on an embedded database within container orchestration environments such as Kubernetes. Doing so can lead to severe data corruption.
Ongoing performance analysis has shown that OrientDB and H2 databases are unsuitable for use beyond the limits below. If you need to exceed the below limits, you should migrate to a PostgreSQL database.
For H2, container-based deployments are not supported. H2 supported workload limitation is 20K requests per day and 100K components; workload beyond these limitations is not supported.
For OrientDB, we also recommend not exceeding 20K requests per day and 100K components.
The temporary directory at
$data-dir/tmp must not be mounted with
noexec or repository manager startup will fail with
java.lang.UnsatisfiedLinkError message of
failed to map segment from shared object: Operation not permitted.
Application Directory - The size of this directory varies slightly each release. It currently around 330 MB. It is normal to have multiple application directories installed on the same host over time as repository manager is upgraded.
Data Directory - On first start, repository manager creates the base files needed to operate. The bulk of disk space will be held by your deployed and proxied artifacts, as well as any search indexes. This is highly installation specific, and will be dependent on the repository formats used, the number of artifacts stored, the size of your teams and projects, etc. It's best to plan for a lot though, formats like Docker and Maven can use very large amounts of storage (500GB easily). When available disk space drops below 4GB the database will switch to read-only mode.
Nexus Repository stores multiple kinds of data, with two primary storage requirements:
- Embedded data (H2, OrientDB, Elastic Search) requires very responsive, fast storage, ideally local disk
- Blob storage (component binaries), which requires moderately responsive, high-capacity storage
File system selection should be made bearing both of these in mind.
|File System||Embedded data||Blob Stores||Comment|
|Local storage||Supported||Supported||Local storage is a good choice for both embedded data and binary storage.|
|NFS v4|| Not Recommended|
|Supported||Most common protocol for network attached storage among Nexus Repository deployments. |
NFSv4.1 or higher can be used for the work directory in small lightly loaded installations, but we have found that it does not provide sufficient performance for anything larger. In general it should be avoided for the work directory.
|Amazon EBS||Supported||Supported||EBS is a viable choice for both embedded data and binary storage.|
|Amazon EFS||Unsupported|| Supported|
|EFS isn't sufficiently responsive for embedded data, but is appropriate for binary storage. |
EFS binary storage may not provide necessary throughput for heavy workloads in all configurations.
|Amazon S3||N/A||Supported||S3 semantics aren't suitable for embedded data, but S3 is popular for binary storage.|
|SMB, CIFS||Unsupported||Supported||Problems are common with SMB or CIFS-mounted devices for embedded data.|
|Azure Blob Storage||N/A||Supported||Available for blob storage from Nexus 3.30.0 Pro. For performance reasons, the Azure blob store should be in the same Azure region as the Nexus Repo installation.|
|Azure Files||Unsupported||Supported||Issues with file handles have been observed when accessing embedded data over SMB.|
|Any fully compatible S3 implementation that supports the AWS SDK version 1.12.265 can be used as a blob store. |
Note that performance characteristics may differ from AWS S3; we recommend working with your vendor to ensure sufficient performance for your desired workload.
|NFS v3||Unsupported||Numerous customers have experienced inadequate performance with NFS v3.|
|GlusterFS||Unsupported||Split-brain problems and slow performance are common.|
|FUSE||Unsupported||FUSE based user-space filesystems are known to be unreliable for Nexus Repository.|
File System Optimization
We also have some optimization suggestions to use at your discretion. Also consider the
noatime option for your Nexus Repository work directory mounts and limit the symbolic links used as this will cause increased overhead whenever paths need to be resolved to an absolute file path.
Our general policy is to support the most recent modern browser version for your supported OS at time of NXRM release date.
|Chrome||latest at NXRM release|
|Mozilla||Firefox||latest and ESR at NXRM release|
|Apple||Safari||latest at NXRM release|
|Microsoft||Edge||latest at NXRM release|
|Microsoft||Internet Explorer||No longer supported|