Skip to main content

Troubleshooting & FAQ

Common Issues​

Issue: Cannot Mount SFS​

Symptoms: SFS fails to mount on compute instance

Possible Causes & Solutions:

  1. Incorrect mount path: Verify the mount path is valid and doesn't conflict with existing directories
  2. SFS in error state: Check the SFS status in the dashboard and contact support if needed

Issue: "Disk Full" Errors​

Symptoms: Applications report disk full even though SFS shows available space

Possible Causes & Solutions:

  1. Inode exhaustion: Too many small files can exhaust inodes. Check inode usage with df -i
  2. Reserved space: Some space may be reserved by the file system
  3. Monitoring lag: Dashboard may not reflect real-time usage. Check actual usage with df -h
  4. Need to resize: If truly full, resize the SFS to add more capacity

Issue: Slow Performance​

Symptoms: Read/write operations are slower than expected

Possible Causes & Solutions:

  1. High concurrent access: Reduce the number of simultaneous operations or scale to PFS
  2. Small file operations: Batch small files or optimize access patterns
  3. Network bottleneck: Check network bandwidth and latency
  4. Consider PFS: For HPC workloads, switch to Parallel File-System for better performance

Issue: Resize Failed​

Symptoms: Resize operation doesn't complete successfully

Possible Causes & Solutions:

  1. Active I/O: Wait for heavy I/O operations to complete before resizing
  2. Backend issue: Check system status and contact support if the problem persists

Troubleshooting & FAQ

Frequently Asked Questions​

General Questions​

Q: What is the maximum size for an SFS?
A: You can resize your SFS up to 64,000 GB. Start with a smaller size and scale up as needed.

Q: Can I decrease the size of my SFS?
A: No, downsizing is not supported. You can only increase the storage capacity. Plan your storage needs accordingly before resizing.

Q: How many Instances can access an SFS simultaneously?
A: Multiple Instances can access the same SFS concurrently with full read/write capabilities. There's no strict limit, but performance may vary based on your workload and access patterns.

Q: Can I rename my SFS after creation?
A: Currently, renaming an SFS after creation is not supported. Choose a descriptive name during creation.

Mounting & Access​

Q: Can I specify a custom mount path for my SFS?
A: Yes, you can specify your preferred mount path when attaching the SFS to compute instances, unlike Datasets which use auto-assigned paths.

Q: What happens to data when I detach an SFS from a node?
A: The data remains intact on the SFS. Detaching only unmounts the file system from that specific node. You can re-attach it later without data loss.

Q: Can I access SFS from my local machine?
A: No, SFS is designed for access from TIR compute instances only. It cannot be mounted directly to local machines.

Performance​

Q: Why is my SFS performance slower than expected?
A: Several factors can affect performance:

  • High concurrent access from many nodes
  • Large numbers of small files
  • Network congestion
  • Workload patterns (random vs sequential access)

Consider using PFS for high-performance parallel I/O workloads if SFS doesn't meet your needs.

Q: What's the difference between SFS and PFS?
A: SFS is optimized for balanced shared access across multiple nodes, while PFS is specifically designed for high-performance parallel I/O in HPC workloads. PFS typically offers better performance for large-scale parallel operations.

Billing & Costs​

Q: How am I charged for SFS?
A: You're charged based on the provisioned storage capacity (GB) and the duration the SFS exists, regardless of actual usage. Pricing is typically calculated per GB per hour.

Q: Do I continue to be charged if no nodes are attached?
A: Yes, you're charged for the provisioned capacity even when no nodes are using the SFS. Delete unused file systems to avoid unnecessary charges.

Data Management​

Q: How do I backup my SFS data?
A: Implement regular backups by:

  • Copying data to Datasets (EOS buckets) for long-term storage
  • Using rsync or similar tools to sync to external storage

Q: What happens to my data if I delete the SFS?
A: All data is permanently deleted and cannot be recovered. Always backup important data before deletion.

Q: What happens to my SFS if I delete an instance that has it attached?
A: The SFS will not be deleted. It will only be unmounted from that instance. Your data remains intact and the SFS continues to exist. You can attach it to other instances or re-attach it to a new instance later.

Data Safety Recommendations​

Backup Strategy​

Critical Reminder

SFS deletion is permanent and irreversible. Always maintain backups of important data.

Recommended Approach:

  1. Regular Backups: Schedule automated backups to external storage
  2. Datasets Integration: Copy critical data to Datasets (EOS) for durable storage

Before Deletion​

  1. Verify Backups: Confirm all important data is backed up elsewhere
  2. Check Dependencies: Ensure no active workloads are using the SFS
  3. Detach Nodes: Unmount the SFS from all attached nodes
  4. Final Review: Double-check that you're deleting the correct file system

Security Best Practices​

  • Access Control: Only attach SFS to instances that require access
  • Data Sensitivity: Understand what data you're storing and compliance requirements
  • Regular Audits: Periodically review which nodes have access to each SFS