wsrep Provider Options in 10.2 CS

Overview

The following options can be set in the wsrep_provider_options system variable to control MariaDB Community Cluster.

Option

Description

Dynamic

base_dir

Specifies the data directory

NO

base_host

Global variable for internal use only

NO

base_port

Global variable for internal use only

NO

cert.length_check

No default

NO

cert.log_conflicts

Log details of certification failures

YES

cert.max_length

No default

NO

cert.optimistic_pa

Controls parallel applying of replica actions

YES

dbug

Default: "" (an empty string)

YES

debug

Enable debugging

YES

evs.auto_evict

Defines how many entries the node allows for given a delayed node before triggering the auto eviction protocol

NO

evs.causal_keepalive_period

For developer use only

NO

evs.debug_log_mask

Controls EVS debug logging when wsrep_debug is in use

YES

evs.delay_margin

Defines how long the node allows response times to deviate before adding an entry to the delayed list

NO

evs.delayed_keep_period

Defines how long the node requires a delayed node to remain responsive before it removes an entry from the delayed list

NO

evs.evict

Sets the point in which the cluster triggers manual eviction to a certain node value

NO

evs.inactive_check_period

Defines how frequently the node should check for peer inactivity

NO

evs.inactive_timeout

Defines a hard limit on node inactivity

NO

evs.info_log_mask

Defines additional logging options for the EVS Protocol

NO

evs.install_timeout

Defines the timeout for install message acknowledgments. This parameter replaces evs.consensus_timeout.

YES

evs.join_retrans_period

Defines how often the node retransmits EVS join messages

YES

evs.keepalive_period

Defines how often the node emits keepalive signals

NO

evs.max_install_timeouts

Defines the number of membership install rounds to try before giving up

NO

evs.send_window

Defines the maximum number of packets at a time in replication

YES

evs.stats_report_period

Control period of EVS statistics reporting

NO

evs.suspect_timeout

Defines the inactivity period after which a node is suspected as dead

NO

evs.use_aggregate

Defines whether the node aggregates small packets into one when possible

NO

evs.user_send_window

Defines the maximum number of data packets the node uses at a time in replication

YES

evs.version

Defines the EVS Protocol version

NO

evs.view_forget_timeout

Defines how long the node saves past views from the view history

NO

gcache.debug

NO

gcache.dir

Defines the directory where the writeset cache places its files

NO

gcache.keep_pages_size

Total size of the page storage pages to keep for caching purposes

NO

gcache.mem_size

The maximum size of the malloc() store for setups that have spare RAM

NO

gcache.name

Defines the filename for the writeset cache

NO

gcache.page_size

Size of the page files in page storage

NO

gcache.recover

Specifies whether gcache recovery takes place on node startup

NO

gcache.size

Defines the amount of disk space the node should use in caching writesets

NO

gcomm.thread_prio

Defines the policy and priority for the gcomm thread

NO

gcs.fc_debug

Specifies the number of writesets that should occur before posting debug statistics about replication flow

NO

gcs.fc_factor

Resume replication after recv queue drops below this fraction of gcs.fc_limit

NO

gcs.fc_limit

Pause replication if recv queue exceeds this number of writesets

YES

gcs.fc_master_slave

Deprecated as of Galera 4.10 in favor of gcs.fc_single_primary

NO

gcs.max_packet_size

Sets the maximum packet size. Writesets that exceed the set size will be fragmented

NO

gcs.max_throttle

Specifies how much to throttle replication rate during state transfer (to avoid running out of memory)

NO

gcs.recv_q_hard_limit

Specifies the maximum allowed size of recv queue

NO

gcs.recv_q_soft_limit

Specifies the fraction of gcs.recv_q_hard_limit after which replication rate will be throttled

NO

gcs.sm_dump

gcs.sync_donor

Specifies if the rest of the cluster should keep in sync with the donor

NO

gmcast.group

gmcast.isolate

gmcast.listen_addr

Specifies the address at which Galera Cluster listens to connections from other nodes

NO

gmcast.mcast_addr

Specifies if UDP multicast will be used for replication

NO

gmcast.mcast_port

gmcast.mcast_ttl

Specifies the time to live value for multicast packets

NO

gmcast.mira

gmcast.peer_addr

gmcast.peer_timeout

Sets the connection timeout to initiate message relaying

NO

gmcast.segment

Defines which network segment this node is in

NO

gmcast.time_wait

Specifies the amount of time to wait before allowing peer declared outside of stable view to reconnect

NO

gmcast.version

Used to check which gmcast protocol version is used. For troubleshooting. Should not be implemented in production.

NO

ist.keep_keys

ist.recv_addr

Specifies the address to listen on for Incremental State Transfer (IST)

NO

ist.recv_bind

Defines the address the node binds on for receiving an Incremental State Transfer (IST)

NO

pc.announce_timeout

Cluster joining announcements are sent every 1/2 second for this period of time or less if the other nodes are discovered

NO

pc.bootstrap

If set to TRUE, this is a signal to turn a NON-PRIMARY component into PRIMARY

NO

pc.checksum

Checksum replicated messages

NO

pc.ignore_quorum

Specifies if quorum calculations should be ignored

YES

pc.ignore_sb

Specifies if nodes are allowed to process updates in a split-brain situation

YES

pc.linger

Specifies the period for which the PC protocol waits for the EVS termination

NO

pc.npvo

If set to TRUE, in the case of conflicting primaries, the more recent Primary Component overrides older ones

NO

pc.recovery

When set to TRUE, the node stores the Primary Component state to disk in the gvwstate.dat file

NO

pc.version

Used for troubleshooting to check which pc protocol version is used

NO

pc.wait_prim

If set to TRUE, the node waits for the pc.wait_prim_timeout time period

NO

pc.wait_prim_timeout

Specifies the period of time to wait for a Primary Component

NO

pc.weight

As of version 2.4. sets the node weight for quorum calculation

YES

protonet.backend

Specifies which transport backend to use. Currently only ASIO is supported

NO

protonet.version

Used for troubleshooting purposes to check which transport backend protocol version is used

NO

repl.causal_read_timeout

Specifies maximum time before causal reads timeout

NO

repl.commit_order

Specifies whether to allow Out-Of-Order committing (improves parallel applying performance)

NO

repl.key_format

Sets the hash size to use for key formats (in bytes)

NO

repl.max_ws_size

Sets the maximum size of a writeset in bytes

NO

repl.proto_max

Used only for debugging. Sets the maximum protocol version in replication

NO

signal

NO

socket.checksum

Specifies which checksum algorithm to use on the socket layer

NO

socket.non_blocking

socket.recv_buf_size

Sets the size of the receive buffer that is used on the network sockets between nodes

YES

socket.send_buf_size

Sets the size of the send buffer that is used on the network sockets between nodes

YES

socket.ssl

Explicitly enables TLS usage by the wsrep provider

NO

socket.ssl_ca

Defines the path to the SSL Certificate Authority (CA) file

NO

socket.ssl_cert

Defines the path to the SSL certificate

NO

socket.ssl_cipher

Specifies which symmetric cipher to use. By default SSL library implementation default cipher is used

NO

socket.ssl_compression

Specifies whether to enable compression on SSL connections

NO

socket.ssl_key

Defines the path to the SSL certificate key

NO

socket.ssl_password_file

Defines a password file for use in SSL connections

NO