When initializing a new cluster, the OneFS installation process creates three directories under the clustered file system root, /ifs:
- /ifs/data
- /ifs/home
- /ifs/.ifsvar
Consider the following guidelines for directories under /ifs.
- Avoid modifying permissions on /ifs/.ifsvar (mode 755). Do not use directory for general storage.
- Avoid modifying /ifs/data/Isilon_Support, which is created the first time the isi_gather_info command is run to collect cluster logs.
- Create hidden shares for /ifs and set ACLs appropriately.
In OneFS 9.0 and later, /ifs/.ifsvar is hidden from view in the CLI, utilities, scripts, protocols, and recursive tools. Anything that attempts to access .ifsvar by full or relative path will still succeed. This feature works by omitting /ifs/.ifsvar when listing this one directory. It only effects the ./ifsvar directory and does not prevent protocols from accessing /ifs/.ifsvar. Prior to OneFS 9.0, /ifs/.ifsvar is hidden from view, but not from tools, scripts, or protocols.
Proper directory layout is critical to a successful OneFS disaster recovery plan.
When planning a cluster’s initial directory structure, consider multitenancy, availability, and disaster recovery requirements.
- During a failover event, NFS clients require their exported paths to remain the same to enable accessing the data. The mount entry for any NFS connection must have a consistent mount point so that during failover, you do not have to manually edit the file system table (fstab) or automount entries on all connected clients.
- OneFS balances a cluster’s connections among all the nodes that service external (front-end) connections. Regularly monitor cluster connections (for example, with InsightIQ). If the number of connections frequently approaches the maximum number the node can support, consider adding another node.
- SmartConnect balances incoming network connections across all the configured network interfaces in a SmartConnect Zone or pool with one of several load-balancing policies. The most common of these policies is round robin, which is effective for most workflows. However, understand whether your front-end connections are being evenly distributed, either in count or bandwidth. Monitor front-end connection distribution.
- Regularly monitor cluster usage with InsightIQ, the OneFS WebUI dashboard, or the command-line interface (CLI). When disk space usage reaches 90 percent, Dell strongly recommends adding additional capacity.
- Many cluster configuration settings are global and have cluster-wide effects. Before changing cluster-wide configuration settings, ensure that you fully understand the global settings and their implications. For information about global cluster configuration settings, see the OneFS Administration Guide
- Confirm that remote support functions work correctly through Secure Remote Services or internal email/SNMP notifications. Note that OneFS 9.1 and beyond only support ESRSv3. Clusters running ESRSv2 will be automatically transitioned to v3 on upgrade to OneFS 9.1 or later.
- OneFS 9.5 provides integration with Dell SupportAssist, the next-generation remote connectivity system for transmitting events, logs, and telemetry from a PowerScale cluster to Dell Support. SupportAssist provides a full replacement for ESRS, as well as enabling Dell Support to perform remote diagnosis and remediation of cluster issues. Intended for all customers who can send telemetry off-cluster to Dell over the Internet, SupportAssist integrates ESE into PowerScale OneFS, along with a suite of daemons to allow its use on a distributed system.
- OneFS provides a recommended protection level based on cluster configuration. This suggested protection level strikes the best balance between data protection and storage efficiency. Follow its recommendations.
- Recommend disabling client DNS caching, where possible. To handle client requests properly, SmartConnect requires that clients use the latest DNS entries. If clients cache SmartConnect DNS information, they might connect to incorrect SmartConnect zone names. In this situation, SmartConnect might not appear to be functioning properly.
- Use LACP on interfaces that carry clients when LACP can be configured across two separate switches to provide switch redundancy.
- Use consistent Ethernet MTU sizes across your network. For example, if using jumbo frames, ensure MTU 9000 is enabled uniformly across the network infrastructure to prevent packet fragmentation or dropped data.
- If round-robin is used as the SmartConnect load balancing strategy with only a small number of high-throughput clients (fewer than 10), the distribution of SMB/NFS connections could result in overloading a few nodes while starving others.