From prior to 2013.3

Follow these instructions if your old version is 2013.2 or earlier.

Important

Compared to earlier versions, the 2013.3 version (and later) contains major changes to the database implementation. These changes allow for increased concurrency and scalability, and removes the 16TB size limit for db.* database files.
Although the db.* database file format has changed, the checkpoint and journal file formats are identical. To upgrade from 2013.2 or earlier to version 2013.3 or later, you must restore the database from a checkpoint.
To upgrade Helix Core Server to a newer version, your license file must be current.
Before shutting down a replica and commencing the upgrade, wait for all archive transfers to end.

To restore from a checkpoint

Stop the Perforce service (p4 admin stop).
Make a checkpoint and back up your old installation. (see Backup procedure)
If a file called tiny.db exists in your old server root, you must back it up separately by running the following command with the old p4d:

p4d -xf 857 > tiny.ckp
Remove the old db.* files, or preferably, move them to a safe location in the event that the upgrade fails by using mv (Linux) or move (Windows). For example:

mv your_root_dir /db.* /tmp

There must be no db.* files in the P4ROOT directory when you rebuild a database from a checkpoint. Although the old db.* files will not be used again, it’s good practice not to delete them until you’re certain your upgrade was successful.
Remove the rdb.lbr file, if it exists.

The rdb.lbr file keeps track of files that need to be transferred to the (local) replica, and might become out of date while the upgrade is underway. Note that this file only exists if your Perforce service was configured as a replica.
Replace the old (2013.2 or earlier) p4d executable with the new (2013.3 or later) p4d executable.

Do not run p4d -xu after replacing p4d at this time. In this upgrade scenario, you are not upgrading an existing database, you have removed it completely and will rebuild it from the checkpoint that you just took.
Use the upgraded p4d to replay the checkpoint and rebuild the new database tables:

p4d -r your/P4ROOT/directory -jr checkpoint_file
If your site uses localized server messages from a message file obtained through Perforce technical support, retrieve the original message.txt file and re-create db.message in the new database format by running the following command with the new p4d:

p4d -jr absolute/path/to/message.txt

See Localizing server error messages for more information.
If you created a tiny.ckp file as part of your backup process, restore tiny.db by running the following command with the new p4d:
```
p4d -xf 857 tiny.ckp
```
Follow the steps at From 2013.3 - 2018.2 to latest (single server) or From 2013.3 - 2018.2 to latest (multi-server environment)
but skip the following steps in those pages:

Shut down the server by running p4 admin stop

Install the new version of the p4d binary on the system.

because you already performed Steps 1 and 6 on this page.