Database Options

Database options and database process options control the activity of database processes. Options specified on the database will be applied to all processes started for that database while database process options apply to Transaction Engine (TE), and Storage Manager (SM) database processes only. All options take effect at process start time for a TE or SM.

The following table provides information for database options. For each option, the table below lists the following:

  • Option Name

  • Option Argument

  • Default Argument (where a default exists)

  • Description of the behavior controlled by the option

  • Restrictions
    Some options can only be applied at the database level when the database is created. These will be used when any processes are started for the database. Some options apply to only certain database processes, for example, the journal-max-directory-entries option applies to SMs but not TEs. These are called database process options. Some options can be specified at the database level or specified as a database process option when the process is started. This distinction is listed in the restrictions column in the table below as one of the following:

    • Database for a database option

    • TE for a database process option that applies to TEs

    • SM for a database process option that applies to SMs

If no restrictions are specified for an option, then it can be used as both database option and database process option.
All options described in the following table are case-sensitive.
Option Name Option Argument Default Argument Description Restrictions

archive-threads

n

1

Number of parallel archive write threads. Increasing the value is recommended when using high latency storage, which is typical with Cloud storage or a Storage Area Network (SAN).

SM

cipher-suites

{TLS|LEGACY-SRP|TLS,LEGACY-SRP}

TLS,LEGACY-SRP

Specifies accepted encryption protocols available for all connections to engines, administration processes and SQL clients. Protocol preference, is defined by the order in which they are listed as comma-separated values. For more information on encryption protocols, see Network Encryption.

commit

{safe|local|remote[:n]|region[:n]}

safe

Commit mode

Database

config

config_file

None

Specifies a configuration file to be passed to the TE, SM process. The configuration file can contain any of the options in this table, represented as key/value pairs, where each key is separated from its value with whitespace, and each key/value pair resides on its own line in the file. This configuration file will be passed to all processes (TEs, SMs) in the database. If the configuration file contains any options that might refer to a local directory path, this path must exist on all database hosts.

crash-dir

directory

$NUODB_LOGDIR/crash

Specifies a directory in which to place crash reports for NuoDB processes (TEs, SMs). This option is for Linux only. $NUODB_LOGDIR for package installations is /var/log/nuodb and for Tar/Local installations is $NUODB_HOME/var/log.

disable-stats-collection

None

None

Disables automatic collection of statistics for indexes, which is enabled by default. If automatic statistic collection is disabled, it is necessary to manually re-calculate index statistics using the ANALYZE command on every table. For more information, see ANALYZE.

enable-query-termination

None

None

Allows TEs to terminate user queries in order to relieve memory pressure. For more information, see SQL Query Termination.

enabled-ciphers

Comma-separated list of cipher names

AES-256-CTR,RC4

A list of acceptable ciphers for clients to connect to the database with. The first cipher is the most preferred. Cipher names are case-sensitive and must be entered exactly as shown. The client will supply a list if ciphers it accepts: if no overlap is found the connection will be refused. The full list of available ciphers are: AES-256-CTR, RC4, and None.

execution-engine

{scalar|vee}

vee

Specifies the default execution engine that each connection will use. This can be overridden on a per-connection basis using the SQLEngine connection-property.

NOTE: If the TE execution-engine property and the connection-property values do not match, then the first connection to the TE using a connection property not matching the execution-engine property will take slightly longer.

TE

hot-copy-threads

n

5

Number of threads used when copying atoms during a hot copy operation. For more information on performing hot copy operations, see Backing Up and Restoring Databases.

SM

io-warn-time

milliseconds

10000
(10 seconds)

A message is logged when a disk I/O operation takes longer than the value specified for this option. These log message are prepended with "Slow I/O:…​".

io-fatal-time

milliseconds

3600000
(1 hour)

A message is logged, and the process is killed, when a disk I/O operation takes longer than the value specified for this option. This avoids a scenario where the entire database hangs because one disk is not responding.

journal-max-directory-entries

n

1000

Sets the maximum number of journal files in a journal sub-directory. A new journal sub-directory is created when the specified limit has been reached.

SM

journal-max-file-size-bytes

n

4000000

Specifies the maximum size in bytes of a journal file. However, the journal may write a file larger than the maximum file size if a single compressed journal message is larger than the specified maximum. The journal messages will be appended to the same journal file until the maximum file size is reached.

SM

journal-single-file

enable/disable

disable

Enables or disables the starting of a new journal file for every sync.

SM

journal-sync-method

{kernel|disk|osync}

disk

Specifies the file system synchronization mechanism when the journal needs to ensure the durability of a commit. The synchronization mechanism has been specifically tuned for each supported OS.

* The kernel synchronization method is the fastest but requires a battery backed disk controller. The kernel method on Linux also requires the file system to support fallocate, that is, ext4. * The disk synchronization method is the most durable but slowest method. * The osync method is available for copy-on-write files systems, also known as ZFS.
For additional information, see Tuning Tips for Journal Performance.

SM

jvm-lib

path

If a value is set for jvm-lib, that location is searched first. If jvm-lib is not set, or a JVM is not found in that location, then JAVA_HOME is searched. If JAVA_HOME is not set or a JVM is not found, then:

On Linux: standard paths are searched.

On Windows: the Windows registry is checked and then common paths are checked.

TE

ldap-base

ldapBase

Required for LDAP authentication. Starting point for the LDAP search instead of the default

Example:

ldap-dn 'ou=people,dc=example,dc=com'

TE

ldap-ca

ldapCA

Optional. LDAP certificate authority certificate to validate server certificate for ldaps. (Linux only)

TE

ldap-conf

ldapConf

Optional. LDAP path to ldap.conf file too use (see man ldap.conf) (Linux only)

TE

ldap-dn

ldapDN

Required for LDAP authentication. This, in conjunction with ldap-pass, allows proxy level authentication and the specified account must have search privileges on the ldap-base. It is used to search for the full Distinguished Name (DN) of the user that connects.

Example:

ldap-dn 'cn=manager, ou=people, dc=example, dc=com

TE

ldap-lib

ldapLib

Optional. LDAP openldap library name (Linux only)

TE

ldap-pass

ldapPassword

Required for LDAP authentication. This, in conjunction with ldap-dn, allows proxy level authentication and the specified account must have search privileges on the ldap-base. It is used to search for the full Distinguished Name (DN) of the user that connects.

TE

ldap-type

unix | windows

Required for LDAP authentication. LDAP server type. Only unix or windows is allowed.

TE

ldap-uri

ldapURI

Required for LDAP authentication. Specify URI referring to

Example:

'ldap://localhost:389'

the LDAP server(s).

TE

max-client-connections

n

0

The number of concurrent client connections a TE will accept.

When starting a TE, use this property to set a specific number of client connections. For more information, see Setting Client Connection Limits for Transaction Engines.

The default (0) allows the TE to accept as many connections as are allowed by the operating system (that is, the same number of sockets that the process can open), minus some reserved for database operations. If the TE cannot accept the number of connections requested (because the operating system does not allow that many open sockets), the TE does not start.

NOTE: Connections are not created until needed by a client, regardless of this setting. NuoDB only checks the max-client-connections value against the operating system’s largest allowed value.

Setting a non-zero value for this startup parameter results in the TE refusing new client connections if the number of current open connections is equal to the value.

TE

max-full-gc-skips

n

100

Specifies the number of normal garbage collection cycles to every full garbage collection cycle (which is every 1000 seconds). By default, there are 100 normal garbage collection cycles to every full garbage collection cycle. For information on generating index statistics, see ANALYZE.

max-http-connections

n

16

Maximum number of concurrent HTTP requests permitted towards the same server. This option is useful when you want to the limit the number of current connections between the SM and the archive. This option can be used only with HDFS and Amazon S3.

TE

max-lost-archives

n

0

The number of leader candidates that may be permanently lost (crash and become non-restartable) at a time before durability is compromised. For a database to be available to commit update transactions, max-lost-archives + 1 leader candidates must be running in the system. The default is 0, meaning that the system is resilient to any number of ordinary failures and available for updates as long as at least one leader candidate is running; but, if 1 or more leader candidates is permanently lost then durability may be compromised.

When max-lost-archives > 0, the transaction can fail to commit, and then roll back. But if the whole database fails before the transaction completes roll back, the partially rolled back transaction can be dredged as pre-committed and promoted to committed, thus creating a torn transaction.

mem

n[K|M|G][i]

2Gi

Sets the maximum amount of memory to be utilized by a database process (in bytes). To specify the amount in kilo-, mega-, giga-, kibi-, mebi-, or gibibytes, suffix the amount with K, M, G, Ki, Mi, or Gi (respectively).

node-port

[n[,n]]

NuoDB decides

Port to use for this host, or optionally, a range of port values. By default NuoDB determines the port to be used.

peer-ciphers

Comma-separated list of cipher names

AES-256-CTR,RC4

A list of acceptable ciphers for peers to connect to the database with. This includes other SMs or TEs as well as connections to the admin services. The first cipher is the most preferred. Cipher names are case-sensitive and must be entered exactly as shown. The client will supply a list if ciphers it accepts: if no overlap is found the connection is refused. The full list of available ciphers are: AES-256-CTR, RC4, and None.

ping-timeout

n

Not set

Normally, processes in the database (TEs, SMs) ping other processes at regular intervals and wait for a response. When network communication with a process times out, the process is evicted from the database by its peers.
Failure detection can be enabled by setting the ping-timeout option, specified in seconds. NuoDB recommends an interval timeout of 60 seconds or greater.
A setting of 0 means wait forever. That is, there is no time out.

Database

remote-syslog

log-options

None

Equivalent to the syslog option but for use with the remote-syslog-server option. The log options you can specify include those listed in the table at the end of this topic.

remote-syslog-server

hostname

Not set

Send all logging information to the server log at the specified remote host.

scratch-dir

directory

N/A

Specifies a directory that NuoDB can use for temporary files to facilitate the processing of memory intensive operations.

For SMs, the default value for --scratch-dir is {archive_dir}/nuodbtemp. For TEs, there is no default value.

TE

scratch-dir-max-size

n[K|M|G][i]

0

Sets the maximum size of the scratch-dir (in bytes). To specify the amount in kilo-, mega-, giga-, kibi-, mebi-, or gibibytes, suffix the amount with K, M, G, Ki, Mi, or Gi (respectively).

0 (the default) means that no maximum size of the scratch-dir is enforced by NuoDB.

stats-collection-threshold

n

10000000000

If any table reaches this cardinality (number of rows), index statistics will cease to be collected automatically for it. Statistics may still be collected manually by issuing the ANALYZE command on the table or any of its indexes. If ANALYZE is issued on the table itself while it is below this cardinality, automatic collection will resume. For more information, see Running ANALYZE to Obtain Index Statistics.

sql-client-tcp-keepalive

tcpKeepalive

Not set

Enable and configure TCP keepalive on incoming SQL client connections. Format tcpKeepalive TCP_KEEPIDLE:TCP_KEEPCNT:TCP_KEEPINTVL. Example: '60:6:10'.

syslog

log-options

Send logging information to the system log. The log options you can specify are listed after this table.

verbose

log-options

None

Report logging and other information. This information is captured in the nuoadmin.conf log file.
You must specify at least one of the log options specified after this table.

To use the remote-syslog, syslog or verbose options, you must specify at least one of the following log options. A log option specifies a log severity level (info, warn, error, debug and flush) or a log category (see Description of Logging Categories).

archive

i18n

security

atoms

index

services

bootstrap

info

sql

client-msgs

journal

sql-index-condition-filtering

debug

msgs

sql-params

dump

net

sql-statements

error

opt-index-selection

stats

failure

opt-joins

table-events

filter

ping

threads

flush

records

transactions

gc

registry

tx

histograms

scheduler

validate

hotcopy

warn