Driver `node`

Supported keywords:

allowed_networks
asset_env
branch
collector
collector_feeder
collector_server
collector_timeout
dbcompliance
dbinsecure
dblog
dbopensvc
env
loc_addr
loc_building
loc_city
loc_country
loc_floor
loc_rack
loc_room
loc_zip
maintenance_grace_period
max_key_size
max_parallel
max_peers
min_avail_mem_pct
min_avail_swap_pct
ready_period
rejoin_grace_period
repo
repocomp
repopkg
ruser
sec_zone
secure_fetch
split_action
sshkey
team_integ
team_support

Keyword `allowed_networks`

required:    false
scopable:    false
default:     10.0.0.0/8 172.16.0.0/24 192.168.0.0/16
convert:     list

Description:

The list of cidr blocks the agents allows creation of backend network into.

Should be restricted to match your site constraints.

Keyword `asset_env`

required:    false
scopable:    false

Example:

asset_env=Production

Description:

An asset information to push to the collector on pushasset, overriding the currently stored value.

Keyword `branch`

required:    false
scopable:    false

Example:

branch=1.9

Description:

Set the targeted opensvc agent branch.

The downloaded upgrades will honor that branch.

If not set, the repopkg imposes the target branch via the current link.

It is recommended to set branch when repopkg points to a repository you are not responsible for.

The system enables OpenSVC Collector 3 calls conditional upon the detection of the Collector 3 environment. If collector_feeder or collector_server are not explicitly defined, they are derived from this value.

Keyword `collector_feeder`

required:    false
scopable:    false

Default:

node.collector/feeder if node.collector is defined. Example:

collector_feeder=https://collector.opensvc.com/feeder

Description:

OpenSVC enables Collector v3 feeder calls upon detection of a collector v3 instance.

Keyword `collector_server`

required:    false
scopable:    false

Default:

node.collector/server if node.collector is defined. Example:

collector_server=https://collector.opensvc.com/server

Description:

OpenSVC enables Collector v3 server calls upon detection of a collector v3 instance.

Keyword `collector_timeout`

required:    false
scopable:    false
default:     5s
convert:     duration

Description:

The maximum time to wait for a collector v3 call. Maximum allowed value 20.

Keyword `dbcompliance`

required:    false
scopable:    false

Default:

Same protocol, server and port as dbopensvc, but with a different path.

Example:

dbcompliance=https://collector.opensvc.com

Description:

Set the uri of the collector’s main rpc server.

The path part of the uri can be left unspecified.

Keyword `dbinsecure`

required:    false
scopable:    false
convert:     bool

Description:

Set to true to disable the collector x509 certificate verification.

This should only be used for testing.

Keyword `dblog`

required:    false
scopable:    false
default:     true
convert:     bool

Description:

If true and dbopensvc is set, the objects action logs are reported to the collector.

Set to false to disable log reporting to the collector, even if dbopensvc is set.

Keyword `dbopensvc`

required:    false
scopable:    false

Example:

dbopensvc=https://collector.opensvc.com

Description:

Set the uri of the collector’s feed rpc server.

The path part of the uri can be left unspecified.

If dbopensvc is not set, the agent does not try to communicate with a collector.

Keyword `env`

required:    false
scopable:    false
default:     TST

Description:

A code like PRD, DEV, etc… the agent can use to enforce data protection policies:

A non-PRD object instance can not be started on a PRD node
A PRD object instance can be started on a non-PRD node (typically in a DRP situation)

Keyword `loc_addr`

required:    false
scopable:    false

Example:

loc_addr=7 rue blanche

Description:

An asset information to push to the collector on pushasset, overriding the currently stored value.

Keyword `loc_building`

required:    false
scopable:    false

Example:

loc_building=Crystal

Description:

An asset information to push to the collector on pushasset, overriding the currently stored value.

Keyword `loc_city`

required:    false
scopable:    false

Example:

loc_city=Paris

Description:

An asset information to push to the collector on pushasset, overriding the currently stored value.

Keyword `loc_country`

required:    false
scopable:    false

Example:

loc_country=fr

Description:

An asset information to push to the collector on pushasset, overriding the currently stored value.

Keyword `loc_floor`

required:    false
scopable:    false

Example:

loc_floor=21

Description:

An asset information to push to the collector on pushasset, overriding the currently stored value.

Keyword `loc_rack`

required:    false
scopable:    false

Example:

loc_rack=R42

Description:

An asset information to push to the collector on pushasset, overriding the currently stored value.

Keyword `loc_room`

required:    false
scopable:    false

Example:

loc_room=102

Description:

An asset information to push to the collector on pushasset, overriding the currently stored value.

Keyword `loc_zip`

required:    false
scopable:    false

Example:

loc_zip=75017

Description:

An asset information to push to the collector on pushasset, overriding the currently stored value.

Keyword `maintenance_grace_period`

required:    false
scopable:    false
default:     60
convert:     duration

Description:

A duration expression, like 1m30s, defining how long the daemon keeps remote node data while it is known to be in maintenance.

The maintenance state is announced to peers at the beginning of a daemon stop and daemon restart, but not on daemon shutdown.

As long as the remote node data is kept, the local daemon won’t takeover the instances running on the node in maintenance.

This parameter should be adjusted to span the daemon restart time.

Keyword `max_key_size`

required:    false
scopable:    false
default:     1mb
convert:     size

Description:

Keyword `max_parallel`

required:    false
scopable:    false
default:     10
convert:     int

Description:

Allow a maximum of max_parallel CRM commands to run simultaneously.

Applies to both:

om <selector> <action> commands.
commands executed by the daemon for orchestrations

Keyword `max_peers`

required:    false
scopable:    false
convert:     int

Example:

max_peers=8

Description:

Keyword `min_avail_mem_pct`

required:    false
scopable:    false
default:     2
convert:     int

Description:

The minimum required available memory to allow orchestration.

Keyword `min_avail_swap_pct`

required:    false
scopable:    false
default:     10
convert:     int

Description:

The minimum required available swap to allow orchestration.

Keyword `ready_period`

required:    false
scopable:    false
default:     5s
convert:     duration

Description:

A duration expression, like 10s, defining how long the daemon waits before starting a service instance in ready state.

A peer node can preempt the start during this period.

Usually set to allow at least a couple of heartbeats to be received.

Keyword `rejoin_grace_period`

required:    false
scopable:    false
default:     90s
convert:     duration

Description:

A duration expression, like 1m30s, defining how long a starting daemon waits in rejoin state.

The daemon normally exits the rejoin state when it has received a heartbeat from all its peer nodes.

During this phase, the orchestration is not allowed, to give a chance to place the services optimally when multiple daemon were restarted at the same time.

But if a peer stays down, the other daemons have to stop waiting at some point to let the service start, even if not on their natural placement leader.

This should be adjusted to:

2s + <longest reboot duration>

The worse case of multiple nodes reboot is when the longest reboot node is rebooted near the end of the reboot of the second longest rebooting node.

|==========>
    n1 reboot
                |--------------------|
                n1 rejoin_grace_period
                |================>
                n1 in rejoin state
                                  |=====================
                                  n1 in idle state
              |==================>
              n2 reboot
                                  |--------------------|
                                  n2 rejoin_grace_period
                                  |=====================
                                  n2 in idle state

As a consequence, to minimize the rejoin_grace_period, prefer fast boot nodes.

Keyword `repo`

required:    false
scopable:    false

Example:

repo=http://opensvc.repo.corp

Description:

Set the uri of the opensvc agent package repository and compliance modules gzipped tarball repository.

This parameter is used by the om node updatepkg and om node updatecomp commands.

Expected repository structure:

Keyword `repocomp`

required:    false
scopable:    false

Example:

repocomp=http://compliance.repo.corp

Description:

Set the uri of the opensvc compliance modules repository.

A gzipped tarball is expected to be found there by the om node updatecomp command.

Expected repository structure:

ROOT
+- compliance-100.tar.gz
+- compliance-101.tar.gz
`- current -> compliance-101.tar.gz

Keyword `repopkg`

required:    false
scopable:    false

Example:

repopkg=http://repo.opensvc.com

Description:

Set the uri of the opensvc agent package repository.

This parameter is used by the om node updatepkg command.

Expected repository structure:

Keyword `ruser`

required:    false
scopable:    false
default:     root

Example:

ruser=root opensvc@node1

Description:

Set the remote user to use to login to a remote node with ssh and rsync.

The remote user must have the privileges to run as root the following commands on the remote node:

om
rsync

The default ruser is root for all nodes.

ruser accepts a list of user[@node]. If @node is omitted, user is considered the new default user.

Keyword `sec_zone`

required:    false
scopable:    false

Example:

sec_zone=dmz1

Description:

An asset information to push to the collector on pushasset, overriding the currently stored value.

Keyword `secure_fetch`

required:    false
scopable:    false
default:     true
convert:     bool

Description:

If set to false, disable ssl authentication checks on all uri fetches.

Keyword `split_action`

required:    false
scopable:    true
candidates:  crash, reboot, disabled
default:     crash

Description:

The node suicide method to use when a cluster split occurs and the node does not have the quorum.

This opting-out is meant to avoid double-start situations when the cluster is split.

Possible values are:

crash

Default.
reboot

May be preferred when the node power-on is not easy. No remote access via IPMI or equivalent for example.
disabled

May be used for test or training only (it does nothing).

Keyword `sshkey`

required:    false
scopable:    false
default:     opensvc

Description:

The basename of the ssh public key served by the GET /node/name/:nodename/ssh/key. For example, the opensvc default value serves ~/.ssh/opensvc.pub.

Keyword `team_integ`

required:    false
scopable:    false

Example:

team_integ=TINT

Description:

An asset information to push to the collector on pushasset, overriding the currently stored value.

Keyword `team_support`

required:    false
scopable:    false

Example:

team_support=TSUP

Description:

An asset information to push to the collector on pushasset, overriding the currently stored value.

Keyboard shortcuts

OpenSVC