Helix Core Server (p4d) Reference

Start the Perforce service, perform checkpoint/journaling, or do certain system administration tasks.

Syntax

p4d [ options ]
p4d.exe [ options ]
p4s.exe [ options ]
p4d -j? [-z | -Z] [ args ... ]
p4d [-z] -R service -jr file 
p4d [-z] [-R service | -P server-id] -jd file  
p4d -Fm | Fs

Description

p4d [ options ]
p4d.exe [ options ]
p4s.exe [ options ]
invoke the background process that manages the Helix Server versioning service
p4d -j? [ -z | -Z ] [ args ... ] for certain system administration tasks, including some that are related to checkpointing and journaling. See Checkpoint files.
p4d -Fm | Fs Failback options
Note

Rotating the journal means saving the existing journal and creating a new, empty journal for future transactions.

"Truncating" a journal refers to the new journal file starting out as an empty file.

On UNIX and macOS, the executable is p4d.

On Windows, the executable is p4d.exe (running as a server) or p4s.exe (running as a service).

These commands should be run during quiet periods

Important

Run the following commands during production downtime or quiet periods. These commands lock tables and therefore might block user commands.

Option Table locking behavior
-jd, -jc, -jcp All tables are locked while the command runs and are unlocked as it completes.
-xu Tables are locked and unlocked one at a time as preparation for the upgrade is done.
-xv Tables are locked and unlocked one at a time as they are being validated.
-xx Tables are locked and unlocked as they are being checked against each other. More than two tables might be locked. For example, the check for db.have against db.rev also locks db.revtx
-xvU Tables are locked and unlocked as they are being checked.

Exit Status

After successful startup, p4d does not normally exit. It merely outputs the following startup message:

Perforce server starting...

and runs in the background.

On failed startup, p4d returns a nonzero error code.

Also, if invoked with any of the -j checkpointing or journaling options, p4d exits with a nonzero error code if any error occurs.

Options

This section includes the following types of options: Server options, General options, Checkpoint and journal options, Journal restore options, Replication and multi-server options, Journal dump and restore filtering, Certificate handling, Configuration options and Failback options.

Server options

Server option Meaning

-d

Run as a daemon (in the background).

-f

Run as a single-threaded (non-forking) process.

-i

Run from inetd on UNIX.

-q

Run quietly (no startup messages).

-n

Start the server in maintenance mode, which means that:

  • the server is only able to perform commands that do not require a client
  • the user and file count restrictions listed in the license file are not enforced

--pid-file[=file]

Write the PID of the server to a file named server.pid in the directory specified by P4ROOT, or write the PID to the file specified by file. This makes it easier to identify a server instance among many.

The file parameter can be a complete path specification. The file does not have to reside in P4ROOT.

--daemonsafe

Is like -d and forks the p4d into the background, but also closes the stdio (standard input output) files.

-xi

Irreversibly reconfigure the Helix Core Server (and its metadata) to operate in Unicode mode. Do not use this option unless you know you require Unicode mode. For details, see the Release Notes and the Internationalization Notes.

-xj

Directly dump the db.ckphist table (see the schema of the Helix Core Server). This option is similar to the p4 journals command. The use case for -xj is to allow you to create a script that determines when a particular checkpoint has completed because it displays the db.ckphist table. For an example of such a script, see Poll for the ckphist record under How to know the parallel checkpoint has completed.

-xj [--limit num] [--jfield flist] [--fieldname fieldvalue]

display checkpoint history records from db.ckphist

--limit restricts the output to the first num entries

--jfield selects the record fields to be displayed from flist

flist is a comma separated list of field names. Valid field names are:

start, startDate, end, endDate, pid, type, flags, jnum, jfile, jdate, jdateDate, jdigest, jsize, jtype, failed, errmsg

Note that the failed and errmsg fields are only present in failed operations.

--fieldname fieldname is one of the valid record fields, such as --jnum or --type

Select only those records where the fieldname field matches fieldvalue

A jnum value of 0 matches any record with an unset jnum field (displayed as -1).

The special fieldname of result with a value of pass selects only successful operations, whereas a value of failed selects only unsuccessful operations.

-xu

Run database upgrades and quit.

  • Upgrades must be run manually unless the server is a DVCS personal server, which runs upgrade steps automatically.

  • Prepare to run a later major release of the server. See Upgrading the server.

  • Tables are locked and unlocked one at a time as preparation for the upgrade is done.

-xv

Run low-level database validation and quit.

  • With no table arguments, validates all tables.

  • If one or more table arguments are provided, only the specified tables are validated. For example:
    p4d -xv db.rev db.change db.have

  • Tables are locked and unlocked one at a time as they are being validated. See also p4 verify in Helix Core Command-Line (P4) Reference.

-xx [-q -o output] [table1 [table2]]

Check consistency between table pairs, producing a jnl.fix file.

  • To restrict output to show only inconsistent tables, use the -q option.

  • If table names are specified, only those tables are checked.

  • Check the jnl.fix file for accuracy before attempting any updates. To specify a filename other than jnl.fix, use the -o option.

  • Tables are locked and unlocked as they are being checked against each other. More than two tables might be locked.

Note

The db.integed/db.rev table pair appears twice in the command output because this table pair check is done in both the toFile direction and in the fromFile direction.

Warning

Do not attempt to replay the jnl.fix file into your depot without consulting Perforce Support.

-xvU

Run fast verification. Does not lock database tables, and verifies only that the unlock count for each table is zero. See also p4 verify in Helix Core Command-Line (P4) Reference.

-xD [serverID]

Display (or set) the server’s serverID (stored in the server.id file) and exit.

-xU CleanServerLocks Safely cleans up the server locks directory, which can help avoid the possible issue of running out of inodes. (See also the Support Knowledgebase article about possible causes of the message No space left on device.)

General options

General option Meaning

-h, -?

Print help message.

-V

Print version number, and other information, such as which memory manager the server uses.

-A auditlog

Specify an audit log file. Overrides P4AUDIT setting. Default is null.

-Id description

A server description displayed by the p4 -z tag info command. Overrides P4DESCRIPTION setting.

-In name

A server name for use with p4 configure. Overrides P4NAME setting. Although this is supported for backward compatibility, an easier method for configuring the server is to use p4 serverid and p4 server.

-J journal

Specify a journal file. Overrides P4JOURNAL setting. Default is journal. (Use -J off to disable journaling.)

-L log

Specify a log file. Overrides P4LOG setting. Default is STDERR.

-p port

Specify a port to listen to. Overrides P4PORT. Default 1666.

-r root

Specify the server root directory. Overrides P4ROOT. Default is current working directory.

-v subsystem=level

Set debug level. Overrides value P4DEBUG setting. Default is null.

-C1

Force the service to operate in case-insensitive mode on a normally case-sensitive platform.

--pid-file[=name]

Write the server’s PID to the specified file.

Default name for the file is server.pid.

--show-realtime Display Monitoring the server in real-time monitoring on the command line. For details of this feature, see p4 monitor realtime in Helix Core Command-Line (P4) Reference.

Checkpoint and journal options

We assume you have read the Backup procedure topic.

To improve performance, consider Parallel checkpointing, dumping and recovery, which is available in checkpoint options that support parallel threads: -jcp, -jcpm, -jdp, -jdpm, and -jvp

Checkpointing option Meaning

-c command

Lock database tables, run command, unlock the tables, and exit.

-jc [-Z | -z] [ prefix ]

Create checkpoint and rotate the journal.

  • All tables are locked while the command runs and are unlocked as it completes.

  • Rotating the journal means saving the existing journal and creating a new journal for future transactions.

p4d -jc creates a new checkpoint.n and rotates the current journal to a numbered journal that corresponds to the now previous checkpoint, journal.n-1

In this case, your checkpoint and journal files are named prefix.ckp.n and prefix.jnl.n respectively, where prefix is as specified on the command line and n is a sequence number.

If no prefix is specified, the default file names checkpoint.n and journal.n are used.

Setting the journalPrefix configurable overrides the default prefix and directory location for rotated journals and checkpoints.

You can store checkpoints and journals in the directory of your choice by specifying the directory as part of the prefix.

With -z, both the checkpoint and the journal are compressed.

With -Z, only the checkpoint is compressed. The journal is left uncompressed.

Without -z or -Z, the checkpoint and journal files are not compressed.

-jcp [-Z | -z] [prefix]
or
-jcp [-N numberOfThreads] [-Z | -z] [prefix]
or
-jcpm [-N numberOfThreads] [-Z | -z] [prefix]

Same as the -jc row above, except that including p supports parallel threads, and using -N numberOfThreads overrides the db.checkpoint.threads configurable.

m is the multifile option.

See Parallel checkpointing, dumping and recovery.

See also Checkpoints for database tree rebalancing.

-jd [-z] file

Create checkpoint without saving or rotating the journal.

  • With -z, the checkpoint is compressed.

  • All tables are locked while the command runs and are unlocked as it completes.

-jdp [-N numberOfThreads] [-z] directory
or
-jdpm [-N numberOfThreads] [-z] directory

Same as the -jd row above, except that including p supports parallel threads, and using -N numberOfThreads overrides the db.checkpoint.threads configurable.

m is the multifile option.

See Parallel checkpointing, dumping and recovery.

[-R service | -P server-id] -jd [-z] file

Similar to -jd [-z] file, but create a checkpoint:

If using -R service If using -P server-id
appropriate for a server providing the services specified by the -R service option, such as -R commit-server or -R edge-server with data relevant to the configured server that has the server ID specified using -P, which causes a journal header to be written containing the server ID

All tables are locked while the command runs and are unlocked as it completes.

-jj [-z] [ prefix ]

Rotate the journal, and no checkpointing occurs. Including -z causes the journal to be compressed.

Your journal file is named prefix.jnl.n, where prefix is as specified on the command line and n is a sequence number. If no prefix is specified, the default filename journal.n is used. You can store journals in the directory of your choice by specifying the directory as part of the prefix.

-jv file
or
-jvp [-N numberOfThreads] directory

Verify the integrity of the checkpoint or journal specified by file as follows:

  • Can the checkpoint or journal be read from start to finish?
  • If it is zipped, can it be successfully unzipped?
  • If it has an MD5 file with its MD5, does it match?
  • Does it have the expected header and trailer?

This command does not replay the journal.

Including p supports parallel threads, and using -N numberOfThreads overrides the db.checkpoint.threads configurable. See Parallel checkpointing, dumping and recovery.

-z

Compress (in gzip format) checkpoints and journals.

When you use this option with the -jd option, Helix Server automatically adds the .gz extension to the checkpoint file. So, the command:

p4d -jd -z myCheckpoint

creates two files: myCheckpoint.gz and myCheckpoint.md5.

Warning

If you have downstream replicas, use -Z instead of -z because they cannot read from a compressed journal.

-Z

Compress (in gzip format) checkpoint, but leave journal uncompressed for use by replica servers. That is, it applies to -jc, not -jd.

Note

Can be used when taking a checkpoint that rotates the journal: p4d -jc -Z

The -Z option is not used for recovery: p4d -jr

Journal restore options

Journal restore option Meaning

-jr file

Restore metadata from a checkpoint and/or journal file.

If the journal header in the checkpoint file contains a server ID and the checkpoint file was created by p4d -P serverID -jd file, a server.id file will be created containing that server ID.

If you specify the -r $P4ROOT option on the command line, the -r option must precede the -jr option.

Note

When using a relative path to identify the file, the path is relative to the P4ROOT folder, not the location from which you ran the command.

-jrp [-N numberOfThreads] directory

Same as the row above, except that the p option supports parallel threads.

Using -N numberOfThreads overrides the db.checkpoint.threads configurable.

See Parallel checkpointing, dumping and recovery.

-R service -jr file

Similar to -jr for restoring metadata from a checkpoint and/or journal file, but only restore metadata appropriate for the service type that is specified by the -R service option, such as -R commit-server or -R edge-server

Tip

The -R edge-server option is useful for the initial seeding of edge servers.

If the journal header in the checkpoint file contains a server ID and the checkpoint file was created by p4d -P serverID -jd file, a server.id file will be created containing that server ID.

-jrc file

Journal-restore with integrity-checking. Because this option locks the database, this option is intended only for use by replica servers started with the p4 replicate command.

-jrF file

Allow replaying a checkpoint over an existing database. (Bypass the check done by the -jr option to see if a checkpoint is being replayed into an existing database directory by mistake.)

-b batch -jr file

Read batch lines of journal records, sorting and removing duplicates before updating the database. The default is 5000, but can be set to 1 to force serial processing. This combination of options is intended for use with replica servers started with the p4 replicate command.

-f -jr file

Ignore failures to delete records. This meaning of -f applies only when -jr is present. This combination of options is intended for use with replica servers started with the p4 replicate command. By default, journal restoration halts if record deletion fails.

As with all journal-restore commands, if you specify the -r $P4ROOT option on the command line, the -r option must precede the -jr option.

-m -jr file

Schedule new revisions for replica network transfer. Required only in environments that use p4 pull -u for archived files, but p4 replicate for metadata. Not required in replicated environments based solely on p4 pull.

-s -jr file

Record restored journal records into regular journal, so that the records can be propagated from the server’s journal to any replicas downstream of the server. This combination of options is intended for use in conjunction with Perforce Support.

-S -jr file

Ignores all storage records in the replay file and instead recalculates the net storage reference count change from the revision records. Applies the recalculated reference count at the end of the replay or the end of each transaction. Like the -s option, the -S option cannot be combined with the -m option.

Important

This combination of options impacts the performance of journal replay. It is intended for small fixes under the guidance of Perforce Support.

Replication and multi-server options

Replication and multi-server option Meaning

-a host:port

In multi-server environments, specify an authentication server for licensing and protections data. Overrides P4AUTH setting. Default is null.

-g host:port

In multi-server environments, specify a changelist server from which to obtain changelist numbers. Overrides P4CHANGE setting. Default is null.

-t host:port

For replicas, specify the target (master) server from which to pull data. Overrides P4TARGET setting. Default is null.

-u serviceuser

For replicas, authenticate as the specified serviceuser when communicating with the master. The service user must have a valid ticket before replica operations will succeed.

Journal dump and restore filtering

Journal dump/restore filtering Meaning

-jd file db.table
or
-jdp -N n
or
-jdpm -N n]

Dump db.table by creating a checkpoint file that contains only the data stored in db.table.

This command can also be used with non-journaled tables.

p supports parallel threads, and [-N n] is the number of threads

m is the multifile option

See Parallel checkpointing, dumping and recovery

-k db.table1,db.table2,... -jd file

Dump a set of named tables to a single dump file.

-K db.table1,db.table2,... -jd file

Dump all tables except the named tables to the dump file.

-P serverid -jd file

Specify filter patterns for p4d -jd by specifying a serverid from which to read filters (see p4 help server, or use the older syntax described in p4 help export).

Tip

This option is useful for seeding a filtered replica.

-R service -jd file

Filter tables based on the service specified:

  • edge-server excludes the tables that are specific to edge servers

    Tip

    This option is useful for the initial seeding of edge servers.

  • commit-server includes all tables

-k db.table1,db.table2,... -jr file

Restore from file, including only journal records for the tables named in the list specified by the -k option.

-K db.table1,db.table2,... -jr file

Restore from file, excluding all journal records for the tables named in the list specified by the -K option.

Certificate handling

Certificate Handling Meaning

-Gc

Generate SSL credentials files for the server: create a private key and certificate file in P4SSLDIR, and then exit.

Requires that P4SSLDIR be set to a directory that is owned by the user invoking the command, and that is readable only by that user. If config.txt is present in P4SSLDIR, generate a self-signed certificate with specified characteristics.

-Gf

Display the fingerprint of the server’s public key, and exit.

Administrators can communicate this fingerprint to end users, who can then use the p4 trust command to determine whether or not the fingerprint (of the server to which they happen to be connecting) is accurate.

Configuration options

Configuration option Meaning

-cshow

Display the contents of db.config without starting the service. (That is, run p4 configure show allservers, but without a running service.)

-cset server#var=val

Set a Helix Server configurable without starting the service, optionally specifying the server for which the configurable is to apply. For example,

p4d -r . "-cset replica#track=1"
p4d -r .  "-cset replica#track=1 replica#server=3"

It is best to include the entire variable=value expression in quotation marks.

-cunset server#var

Unset the specified configurable.

Failback options

In regard to Failback after failover, we recommend that you first preview the commands and only include -y if the preview looks acceptable.

Command Purpose

p4d -r p4root -Fm masterServerID standbyServerID

To preview the conversion of the original master server to a standby for the current master.

p4d -r p4root -y -Fm masterServerID standbyServerID

To perform the conversion of the original master server to a standby for the current master. The converted standby will then be ready for a later p4 failback command.

p4d -r p4root -Fs masterServerID standbyServerID

(Optional) To preview the conversion of the original standby server to a standby for the restored master.

p4d -r p4root -y -Fs masterServerID standbyServerID

(Optional) To perform the conversion of the original standby server to a standby for the restored master.

Usage Notes

  • Do not run p4d as root. See Running the Helix Server (p4d) as an unprivileged user.
  • On all systems, journaling is enabled by default. If P4JOURNAL is unset when p4d starts, the default location for the journal is $P4ROOT. If you want to manually disable journaling, you must explicitly set P4JOURNAL to off.
  • Take checkpoints and truncate the journal often, preferably as part of your nightly backup process.
  • Checkpointing and journaling preserve only your Helix Server metadata (data about your stored files). The stored files themselves (the files containing your source code) reside under P4ROOT and must be also be backed up as part of your regular backup procedure.
  • It is best to keep journal files and checkpoints on a different hard drive or network location than the Helix Server database.
  • If your users use triggers, don’t use the -f (non-forking mode) option. To run trigger scripts, the Perforce service needs to be able to "fork" (spawn copies of itself).
  • After a hardware failure, the options required for restoring your metadata from your checkpoint and journal files can vary, depending on whether data was corrupted.
  • Because restorations from backups involving loss of files under P4ROOT often require the journal file, we strongly recommend that the journal file reside on a separate filesystem from P4ROOT. This way, in the event of corruption of the filesystem containing P4ROOT, the journal is likely to remain accessible.
  • The database upgrade option (-xu) can require considerable disk space. For details, see the Release Notes.

Typical tasks

To start the service, listening to port 1999, with journaling enabled and written to journalfile.

p4d -d -p 1999 -J /opt/p4d/journalfile

To checkpoint a server with a non-default journal file, the -J option (or the environment variable P4JOURNAL) must match the journal file specified when the server was started.

Checkpoint with:

p4d -J /p4d/jfile -jc

or

P4JOURNAL=/p4d/jfile ; export P4JOURNAL; p4d -jc

To compress checkpoints and journals, which creates two files: myCheckpoint.gz and myCheckpoint.md5. p4d -jd -z myCheckpoint

To create a compressed checkpoint from a server with files in directory P4ROOT.

p4d -r $P4ROOT -z -jc

To create a compressed checkpoint with a user-specified prefix of “myCheckpoint” from a server with files in directory P4ROOT.

p4d -r $P4ROOT -z -jc myCheckpoint

To restore metadata from a checkpoint named checkpoint.3 for a server with root directory P4ROOT.

p4d -r $P4ROOT -jr checkpoint.3

To restore metadata from a compressed checkpoint named checkpoint.3.gz for a server with root directory P4ROOT.

p4d -r $P4ROOT -z -jr checkpoint.3.gz