Dell NetWorker 19.9 Administration Guide

Improved resilience of long running write operations

To improve the resilience of long running write operations to a Data Domain (DD) device, especially for backups from NFS and NDMP clients, the option, resilient write, is available.

If there is no read/write operation to a DD file handle for the timeout (OST_ABANDON_TIMEOUT for which the default value is three hours) value mentioned, the write operations may get abandoned with a stale file handle error (DD_ERR_STALE). This forces the administrator to restart the backup job, which has to traverse through the filesystem and process the data all over again.

With resilient write enabled, the client can continue to perform write operations and complete the backup process even after encountering the stale file handle error. save and nsrdsa_save binaries perform the fsync operation on the device for every 128 MB and also hold a cache of size 128 MB to store intermittent data before the next fsync operation.

Resilient write for backups can be enabled only for NFS and NDMP clients with client direct enabled. It is developed for backups to DD devices.

By default, resilient write is disabled for all backups jobs. It can be configured for specific clients (NDMP or NFS) using the following methods:

• Configure resilient write for NDMP clients
• Configure resilient write for NFS clients