Skip to content

Latest commit

 

History

History
279 lines (248 loc) · 11.6 KB

server_config.md

File metadata and controls

279 lines (248 loc) · 11.6 KB
Raw

Server configuration

There are couple of sections in configuration file:

API

Options:

  • addr (optional) - Address (without scheme) API server will bind to. If certfile or keyfile is set then HTTPS will be used, otherwise HTTP ("localhost:8000" by default).
  • certfile (optional) - Absolute path to certificate.
  • keyfile (optional) - Absolute path to private key.
  • auth (optional)

    • backend (optional) - "none", "gitlab" or "ldap" ("none" by default).

    • gitlab (optional and used only if backend is set to "gitlab")

      • addr (required) - GitLab address with scheme like https://.
      • cacert (optional) - Absolute path to CA certificate to use when verifying GitLab server certificate, must be x509 PEM encoded.

    • ldap (optional and used only if backend is set to "ldap")

      • addrs (required) - List of LDAP server addresses with scheme ("ldap://" or "ldaps://") and optional port.
      • userdn (required) - Base DN under which to perform user search.
      • userattr (required) - Attribute on user object matching the username passed when authenticating ("cn" by default).
      • binddn (optional) - Distinguished name of object to bind when performing user search.
      • bindpassword (optional) - Password to use along with binddn when performing user search.
      • cacert (optional) - Absolute path to CA certificate to use when verifying LDAP server certificate, must be x509 PEM encoded.
      • timeout (optional) - Connect timeout in milliseconds for single connection to one server (5000 by default).
      • caseSensitiveNames (optional) - If set, LDAP user and group names lookups in useracl and groupacl will be case sensitive. Otherwise, names will be normalized to lower case - use lowe case in groupacl and useracl to make matching work. Case will still be preserved when sending the username to the LDAP server at login time; this is only for matching user/group ACLs. (false by default).
      • groupfilter (optional) - Go template used when constructing the group membership query. The template can access the following context variables: [UserDN, Username]. Use "(&(objectClass=group)(member:1.2.840.113556.1.4.1941:={{.UserDN}}))" to support nested group resolution for Active Directory ("(|(memberUid={{.Username}})(member={{.Us erDN}})(uniqueMember={{.UserDN}}))" by default).
      • groupdn (required) - LDAP search base to use for group membership search like "ou=Groups,dc=example,dc=com". This can be the root containing either groups or users.
      • groupattr (required) - LDAP attribute to follow on objects returned by groupfilter in order to enumerate user group membership. Examples: for groupfilter queries returning group objects, use: "cn". For queries returning user objects, use: "memberOf" ("cn" by default).
      • useracl (optional) - Access control list defining permission level ("readwrite" or "readonly") per LDAP user. Possible to define it for project, the whole group or everything / all projects ("*"). Example:
       "useracl": {
 	"john": {
 		"infra": "readonly",
 		"infra/monitoring": "readwrite"
 	},
 	"katie":  {
 		"infra/monitoring": "readwrite",
 		"*": "readonly"
 	}
 }

      If accesses for both the whole group and for specific project in that group are set then latter is used (john has read-write access to infra/monitoring and read-only access to any other project from infra group).

      Match-all ("*") has the least priority so if set and also group or project is specified then latter is used (katie has read-write access to infra/monitoring and read-only access to any other project).

      See API documentation for explanation which access level is taking into account if both useracl and groupacl are set.

      • groupacl (optional) - Access control list defining permission level ("readwrite" or "readonly") per LDAP group. Possible to define it for project, the whole group or everything / all projects ("*"). Example:
       "groupacl": {
 	"qa": {
 		"infra": "readonly",
 		"infra/monitoring": "readwrite"
 	},
 	"devs":  {
 		"infra/monitoring": "readwrite",
 		"*": "readonly"
 	}
 }

      If accesses for both the whole group and for specific project in that group are set then latter is used (qa LDAP group members have read-write access to infra/monitoring and read-only access to any other project from infra group).

      Match-all ("*") has the least priority so if set and also group or project is specified then latter is used (devs LDAP group members have read-write access to infra/monitoring and read-only access to any other project).

      See API documentation for explanation which access level is taking into account if both useracl and groupacl are set.

Examples:

"api": {
    "addr": "localhost:8888",
    "auth": {
        "backend": "gitlab",
        "gitlab": {
            "addr": "https://example.com",
            "cacert": "/var/ca.crt"
        }
    }
}
"api": {
    "addr": "localhost:8888",
    "auth": {
        "backend": "ldap",
        "ldap": {
            "addrs": ["ldap://example.com"],
            "userdn": "dc=example,dc=com",
            "userattr": "uid"
        }
    }
}

Storage

Options:

  • backend (optional) - "zookeeper" ("zookeeper" by default).
  • zookeeper (optional and used only if backend is set to "zookeeper")
    • dir - Location (name without slashes) to store data ("rhythm" by default).
    • addrs - Servers locations without scheme. If port is not set then default 2181 will be used (["127.0.0.1"] by default).
    • timeout (optional) - ZooKeeper client timeout in milliseconds (10s by default).
    • auth (optional)
      • scheme (optional) - "digest" or "world" ("world" by default).
      • digest (optional and used only if scheme is set to "digest")
        • user (optional)
        • password (optional)
    • taskttl (optional) - number of milliseconds record of runned task should be kept (7 days by default).

Example:

"storage": {
    "zookeeper": {
        "addrs": ["192.168.0.1", "192.168.0.2", "192.168.0.3"],
        "dir": "rhythm",
        "timeout": 20000,
        "auth": {
            "scheme": "digest",
            "digest": {
                "user": "john",
                "password": "secret"
            }
        }
    }
}

Coordinator

Options:

  • backend (optional) - "zookeeper" ("zookeeper" by default).
  • zookeeper (optional and used only if backend is set to "zookeeper")
    • dir - Location (name without slashes) to keep state ("rhythm" by default).
    • addrs - Servers locations without scheme. If port is not set then default 2181 will be used (["127.0.0.1"] by default).
    • timeout (optional) - ZooKeeper client timeout in milliseconds (10s by default).
    • auth (optional)
      • scheme (optional) - "digest" or "world" ("world" by default).
      • digest (optional and used only if scheme is set to "digest")
        • user (optional)
        • password (optional)

Example:

"coordinator": {
    "zookeeper": {
        "addrs": ["192.168.0.1", "192.168.0.2", "192.168.0.3"],
        "dir": "rhythm",
        "timeout": 20000,
        "auth": {
            "scheme": "digest",
            "digest": {
                "user": "john",
                "password": "secret"
            }
        }
    }
}

Secrets

Secrets backend allow to inject secrets into task via environment variables. Job defines secrets under secrets property:

"group": "webservices",
"project": "oauth",
"id": "backup",
"secrets": {
    "DB_PASSWORD": "db/password"
}

Mesos task will have DB_PASSWORD environment variable set to value returned by secrets backend if "webservices/oauth/db/password" will be passed. In case of e.g. Vault it'll be interpreted as path to secret from which data under value key will retrieved.

Options:

  • backend (optional) - "vault" or "none" ("none" by default).
  • vault (optional and used only if backend is set to "vault")
    • addr (required) - Vault address with scheme like https://.
    • token (required) - Vault token with read access to secrets under root.
    • root (optional) - Secret's path prefix ("secret/rhythm/" by defualt).
    • timeout (optional) - Client timeout in milliseconds (0 by default which means no timeout).
    • cacert (optional) - Absolute path to CA certificate to use when verifying Vault server certificate, must be x509 PEM encoded.

Example:

"secrets": {
    "backend": "vault",
    "vault": {
        "token": "aaaaaaaa-aaaa-aaaa-aaaa-aaaaaaaaaaaaa",
        "addr": "https://example.com"
    }
}

Mesos

Options:

  • addrs (required) - List of Mesos endpoints with schemes like https://.
  • auth
    • type (optional) - "none" or "basic" ("none" by default).
    • basic (optional and used only if type is set to "basic")
      • username (optional)
      • password (optional)
  • cacert (optional) - Absolute path to CA certificate to use when verifying Mesos server certificate, must be x509 PEM encoded.
  • checkpoint (optional) - Controls framework's checkpointing (false by default).
  • failovertimeout (optional) - Number of milliseconds Mesos will wait for the framework to failover before killing all its tasks (7 days used by default).
  • hostname (optional) - Host for which framework is registered in the Mesos Web UI.
  • user (optinal) - Determine the Unix user that tasks should be launched as.
  • webuiurl (optional) - Framework's Web UI address with scheme like https://.
  • principal (optional) - Identifier used while interacting with Mesos.
  • labels (optional) - Dictionary of key-value pairs assigned to framework.
  • roles (optional) - List of roles framework will subscribe to (["*"] by default).
  • logallevents (optional) - Print details of all events sent from Mesos (false by default).

Example:

"mesos": {
    "addrs": ["https://example.com:5050"],
    "principal": "rhythm",
    "roles": ["rhythm"],
    "user": "root",
    "webuiurl": "https://example.com",
    "auth": {
        "type": "basic",
        "basic": {
            "username": "rhythm",
            "password": "secret"
        }
    },
    "labels": {
        "one": "1",
        "two": "2"
    }
}

Logging

Logs are always sent to stderr (level defines verbosity) and optional backend to e.g. send certain messages to 3rd party service like Sentry.

Options:

  • level (optional) - "debug", "info", "warn" or "error" ("info" by default).

  • backend (optional) - "sentry" or "none" ("none" by default).

  • sentry (optional and used only if backend is set to "sentry")

    Logs with level set to warning or error will be sent to Sentry. If logging level is higher than warning then only errors will be sent (in other words level defines minium tier which will be by Sentry backend).

    • dsn (required) - Sentry DSN (Data Source Name) passed as string.
    • cacert (optional) - Absolute path to CA certificate to use when verifying Sentry server certificate, must be x509 PEM encoded.
    • tags (optional) - Dictionary of custom tags sent with each event.

Examples:

"logging": {
    "level": "debug",
    "backend": "sentry",
    "sentry": {
        "dsn": "https://[email protected]/123",
        "cacert": "/var/ca.crt",
        "tags": {
            "one": "1",
            "two": "2"
        }
    }
}
"logging": {
    "level": "debug"
}

There is -testlogging option which is used to test events logging. It logs sample error and then program exits. Useful to test backend like Sentry to verify that events are received.