Manage the NFS protocol
The WEKA system enables file access through the NFS protocol instead of the WEKA client.
Last updated
The WEKA system enables file access through the NFS protocol instead of the WEKA client.
Last updated
NFS (Network File System) is a protocol that enables clients to access the WEKA filesystem without requiring WEKA's client software. This leverages the standard NFS implementation of the client's operating system.
WEKA supports an advanced NFS implementation, NFS-W, designed to overcome inherent limitations in the NFS protocol. NFS-W is compatible with NFSv3 or protocols and offers enhanced capabilities, including support for more than 16 user security groups and NFS file-locking.
Adhere to the following guidelines and requirements when deploying the NFS service.
A persistent cluster-wide configuration filesystem is required for the protocol's internal operations using NFSv4 or Kerberos integration. See .
An interface group is a configuration framework designed to optimize resiliency among NFS servers. It enables the seamless migration of IP addresses, known as floating IPs, from an unhealthy server to a healthy one, ensuring continuous and uninterrupted service availability.
An interface group consists of the following:
A collection of WEKA servers with a network port for each server, where all the ports must be associated with the same subnets. For resiliency, a minimum of two NFS servers are required.
A collection of floating IPs to support the NFS protocol on specified servers and NICs. All IP addresses are required to be within the same subnet, and the servers must already have static IP addresses on those NICs within that subnet.
A routing configuration for the IPs. The IP addresses must comply with the IP network configuration.
An interface group can have only a single port. Therefore, two interface groups are required to support High Availability (HA) in NFS. When assigning the other server ports to these interface groups, consider the network topology to ensure no single point of failure exists in the switch.
You can define up to 10 different Interface groups. Use multiple interface groups if the cluster connects to multiple subnets. You can set up to 50 servers in each interface group.
The WEKA system automatically distributes the IP addresses evenly on each server and port. If a server fails, the WEKA system redistributes the IP addresses associated with the failed server to other servers.
The WEKA system automatically configures the floating IP addresses used by the NFS service on the appropriate server. Refrain from manually configuring or using the floating IP.
To ensure load balancing between the NFS clients on the different WEKA servers serving NFS, it is recommended that a round-robin DNS entry be resolved to the list of floating IPs.
Related information
The NFS client mount is configured using the standard NFS stack operating system. The NFS server IP address must point to the round-robin DNS name.
Access Control List (ACL) in NFS (Network File System) provide fine-grained control over file permissions, offering more flexibility than traditional POSIX permissions. NFS supports multiple ACL flavors, each serving different use cases and interoperability needs.
ACL types in NFS
NFS supports the following ACL types:
None: No ACL enforcement or updates occur, even if POSIX ACLs exist on a file or directory. This flavor is used when ACL management is unnecessary.
POSIX: NFS enforces POSIX ACLs, ensuring compatibility with other protocols. However, the finer granularity of NFSv4 ACLs is lost when mapped to POSIX ACLs. This option is suitable for environments requiring basic ACL management across multiple protocols.
NFSv4: NFSv4 ACLs are enforced directly, without mapping to POSIX ACLs. This flavor preserves the full granularity of NFSv4 ACLs but does not support interoperability with other protocols. ACLs are stored as extended attributes and mapped to user and group IDs (UID/GID). Use NFSv4 when full NFSv4 ACL granularity is required, and interoperability with other protocols is not a concern.
Hybrid: This flavor combines both POSIX and NFSv4 ACLs to support interoperability. NFS ensures consistency between the two ACL types, and if any inconsistency arises, POSIX ACL is used for enforcement. Hybrid is ideal for environments requiring both interoperability and full NFSv4 ACL functionality.
NFSv3 and ACLs
NFSv3 supports only the None and POSIX ACL flavors. Any attempt to set ACLs from NFSv3 clients using other flavors is not applicable. NFS does not enforce ACLs on NFSv3 clients, but the underlying filesystem handles access control if POSIX ACLs are present.
Managing ACLs in NFS
ACL configuration and management in NFS can be done through various interfaces:
CLI: The weka nfs permission and weka nfs global-config commands allow users to configure ACLs at the permission and cluster level. NFS permissions exporting files from the same backend file system must share the same ACL flavor.
GUI: The NFS settings in the user interface include options to enable ACLs and configure default ACL flavors (None, POSIX, NFSv4, Hybrid). Changes to ACL settings may require restarting the NFS containers.
Configuration filesystem: ACL flavors and related configurations are tracked in the global configuration filesystem, ensuring consistent management of permissions across the system.
LDAP and ACLs
For ACL functionality, LDAP must be configured to manage user and group information. If Kerberos is set up, no additional LDAP configuration is necessary. Otherwise, LDAP configuration options are provided, supporting both Active Directory (AD) and OpenLDAP.
Upgrading and ACLs
When upgrading, the default ACL flavor for all permissions sets to POSIX. ACLs are enabled by default. To ensure proper ACL functionality, both .config_fs and LDAP must be configured.
WEKA facilitates the seamless integration of NFS with an existing Kerberos service. This integration enables clients' authentication, data integrity, and data privacy over the wire when interacting with the NFS server, ensuring robust security even across untrusted networks.
The Kerberos security levels are:
krb5: Implements basic Kerberos authentication.
krb5i: Incorporates Kerberos authentication with data integrity assurance.
krb5p: Integrates Kerberos authentication with data integrity and privacy measures.
WEKA supports Kerberos authentication for NFS using AD and Kerberos MIT:
Active Directory (AD): NFS integrates with Active Directory (AD), which includes built-in Kerberos services. WEKA interacts with the AD using the Kerberos protocol to authenticate service requests among trusted devices.
Kerberos MIT: NFS integrates with Kerberos MIT, implementing the Kerberos protocol, which uses secret-key cryptography for authentication across insecure networks. This protocol is widely standardized and utilized.
The following Kerberos service interactions ensure secure communication between the client and the WEKA NFS server:
Client authentication and ticket request: The client sends a request, including encrypted credentials, to the Authentication Server for a Ticket Granting Ticket (TGT).
Ticket generation and delivery: The Authentication Server verifies the client’s identity, generates a session key, forms a TGT, and sends these to the client.
Ticket extraction and service request: The client decrypts the received message, extracts the session key and the TGT, and sends a service request to the Ticket Granting Server.
Service session key generation & ticket formation: The Ticket Granting Server verifies the TGT, generates a Service Session Key, and forms a Service Ticket.
Service ticket delivery & extraction: The Ticket Granting Server sends the Service Ticket and the Service Session Key to the client, who then decrypts the response and extracts these for later use.
Service access & verification: The client generates an authenticator for the network service and sends it along with the Service Ticket to the network service, which then verifies the Service Ticket and the authenticator.
Add as many servers as possible to the interface group for performance scalability.
facilitate load balancing by evenly distributing them across all interface group servers and ports, given the system has 50 or fewer NFS interfaces. However, with the limitation of 50 floating IPs per cluster, systems with more than 50 NFS interfaces may not have a floating IP for each interface.
When different clients resolve the DNS name into an IP service, each receives a different IP address, ensuring that other clients access different servers. This allows the WEKA system to scale and service thousands of clients.
To ensure the resilience of the service if a server fails, the system reassigns all IP addresses associated with the failed server to other servers (using the network messages), and the clients reconnect to the new servers without any reconfiguration or service interruption.
NFS file-locking is a mechanism that ensures synchronized access to files by multiple processes in a networked environment. It maintains data integrity and consistency by preventing simultaneous access to the same file, thereby preventing potential data corruption.
Port prerequisites: Ports used by the nlockmgr and status services must be open on the clients and WEKA servers. Use one of the following methods to meet this requirement:
Disable and stop firewalld using the commands:
Define the ports in /etc/services and restart the rpc.statd (ensure the port numbers are open). For example:
NFS client prerequisite: To use NFSv3 with locking on an NFS client, ensure the rpc.statd service runs in the NFS client. This enables clients to mount NFSv3 shares.
To inspect the active locks on a specific file, use the following command:
<inode-id>: The unique identifier of the file’s inode.
--snap-view-id snap-view-id: (Optional) Specifies the snapshot view ID for listing locks on a file within a particular snapshot.
--verbose: (Optional) Provides detailed lock information, including the lock owner and type.
This command outputs a list of all current locks on the specified file, enabling administrators to monitor and manage file access effectively.
For detailed procedures, see the related topics.
Related topics
