Home > Storage > PowerScale (Isilon) > Product Documentation > Storage (general) > PowerScale OneFS: Long Filename Support > SyncIQ requirements for long filenames
The increase in filename length in OneFS 9.3 and later beyond the previous limit of 255 bytes introduces an incompatibility with previous OneFS versions. As such, files that are created with names that exceed 255 bytes cannot be replicated using SyncIQ to a cluster running a OneFS version earlier than 9.3. This incompatibility is identified before SyncIQ attempts to replicate any file from the source cluster to an incompatible target cluster. When the incompatibility is found, the SyncIQ policy fails, a CELOG event is generated, and the replication session terminates without any files being transferred.
If long filename support is enabled on a cluster with active SyncIQ replication policies, all source and target clusters with a remote replication dependency must satisfy the following prerequisites:
A cluster containing long-named files cannot replicate to a cluster that has not been upgraded to OneFS 9.3 or later, so any SyncIQ target clusters must be updated before long name support is enabled on the source cluster. The long filename configuration does not have to be identical between the source and target clusters; it only has to be enabled. SyncIQ will successfully replicate long-named files to a target cluster with a restricted/default policy.
For any cluster not meeting the above-described prerequisites, SyncIQ replication policies will fail upon an attempt to establish a connection. You can enable long filenames on a SyncIQ target cluster either by creating a long filename configuration using the isi namelength create CLI command, as described previously, or by setting sysctl to value 1 in the following long_file_name_enabled syntax. For example:
# sysctl efs.bam.long_file_name_enabled=1”
When the target cluster does not support long filenames for a SyncIQ policy and the source domain has long filenames enabled, the replication job fails. The subsequent SyncIQ job report includes the following error message:
A similar failure report is displayed through the CLI:
# isi sync reports view LFN_test 1
Policy Name: LFN_test
Job ID: 1
Start Time: 2021-05-19T21:11:01
End Time: 2021-05-19T21:11:07
Action: run
State: failed
ID: 1-LFN_test
Policy ID: 7a80761e4b013dcd8141b9ff1a5c1fb7
Sync Type: initial
Duration: 6s
Errors: Cannot sync from a cluster with long file names to a target cluster on a version too old to provide long name support
Encrypted: No
Source Directories Visited: 0
Source Directories Deleted: 0
Quotas Deleted: 0
Num Retransmitted Files: 0
Retransmitted Files: -
Total Files: 0
New Files Replicated: 0
Source Files Deleted: 0
Files Changed: 0
Target Files Deleted: 0
Worm Committed File Conflicts: 0
Committed Files: 0
Up To Date Files Skipped: 0
User Conflict Files Skipped: 0
Error Io Files Skipped: 0
Error Net Files Skipped: 0
Error Checksum Files Skipped: 0
Bytes Transferred: 0
Total Network Bytes: 0
Total Data Bytes: 0
File Data Bytes: 0
Sparse Data Bytes: 0
Target Snapshots: -
Total Phases: 0
Subreports:
Policy Name Job ID Subreport ID Start Time End Time Action State
-----------------------------------------------------------------------------------------
LFN_test 1 1 2021-05-19T21:11:01 2021-05-19T21:11:07 run failed
-----------------------------------------------------------------------------------------
Total: 1
A CELOG event and alert are also generated when a SyncIQ replication policy fails due to the target cluster’s inability to support long filenames:
With SyncIQ cascaded replication configurations, only the immediate target can be verified for long filename compatibility from the source cluster. The OneFS checks that verify and fail an incompatible SyncIQ job cannot identify and terminate instances in which the first-level target is running OneFS 9.3 or later with long filename support enabled, but where the cascaded target is running an earlier OneFS version and/or does not have long filenames configured. In these cases, SyncIQ fails without warning on the cascaded cluster: