VMware SRM offers a wide variety of advanced options and settings that can be altered by the SRM administrator if required, or if requested by Dell or VMware support. Most of these options are well beyond the scope of this paper, but this describes a few key settings involved in the storage configuration and behavior of SRM.
If these settings are changed from the default, it is recommended to make the change on both the protected and recovery SRM servers to maintain configuration consistency between the two environments. Changing an option on one SRM server does not propagate to the other SRM server the user must ensure it is done. Below are descriptions and recommendations for a list of six of the most common storage related advanced SRM settings.
- storage.commandTimeout: This option controls the amount of time SRM allows the SRA to run an operation. In small to medium environments the default of 300 s is sufficient but in larger or more complex environments, or in situations where the SRA must wait for data to synchronize, increasing this setting may be advisable. It is recommended to not adjust this setting unless it is deemed necessary. This option affects both recovery and test recovery operations.
- storage.storagePingInterval: This option controls the time interval between SRM-initiated automatic SRA rediscovery of the storage arrays and replication information. The default setting is to rediscover every 24 hours (configured in seconds in SRM). In general, it is not recommended to change this setting as it is important that SRM and the SRA update regularly to ensure that the environment is in a valid state.
- storageProvider.fixRecoveredDatastoreNames: During recovery or test recovery, a prefix is added to datastore names when the ESXi servers resignature and mount recovered datastores. This option is true or false. When enabled, SRM removes the snap-xxxxxx prefixes that are automatically added to recovered datastore names. SRM removes these prefixes after mounting the volume but before registering any virtual machines. By default, this setting is disabled and datastores keep the prefixes unless manually removed. For customers that run scripts based on datastore name, Dell Technologies recommends enabling the setting to avoid errors. This option affects both recovery and test recovery operations.
- storageProvider.hostRescanDelaySec: This option controls the amount of time between when the PowerFlex SRA returns success to SRM after a failover and when SRM initiates a storage rescan on the ESXi hosts. If the storage devices are not fully available yet, ESXi does not detect them and SRM does not find the replicated devices when it performs rescans. This would prevent the mounting of datastores and registering of the virtual machines. If an increase to this delay from the default of zero is necessary, a value of between 20s to 180s is reasonable. Testing generally surfaces the need to adjust this setting. This option affects both recovery and test recovery operations.
- storageProvider.autoResignatureMode: This option controls the process of resignaturing during recovery of datastores. This option affects both recovery and test recovery operations. There are three different modes for this option:
- Disable: Uses a selective resignature API to query and resignature only the snapshot VMFS volumes relevant to SRM. This is the default and recommended mode. SRM checks before every HBA rescan (during every test and failover operation) to check that the LVM.EnableResignature option on all ESXi hosts is disabled, and if it is not, SRM disables it and leaves it as such. This mode is selected by setting the option value to 0.
- Enable: Turns on the LVM/EnableResignature flag in the respective ESXi hosts to enable automatic resignature of ALL unresolved VMFS volumes. SRM checks before every HBA rescan (during every test and failover operation) to check that the LVM.EnableResignature option on all ESXi hosts is enabled, and if it is not, SRM enables it and leaves it as such. No snapshot is left out regardless of its relationship to SRM. Dell strongly discourages the use of this mode. This mode is selected by setting the option value to 1.
- Ignore: Ignores the LVM/EnableResignature flag on the respective ESXi hosts. Whatever the user has preconfigured in ESXi for LVM/EnableResignature, SRM uses it. SRM checks if volumes are auto resignatured, if not it queries and selectively resignatures them. This mode is selected by setting the option value to 2. Note: Setting the LVM.enableResignature flag on ESXi hosts is a host-wide operation and, if set, all snapshot LUNs that can be resignatured are resignatured during the subsequent host rescan. If snapshot volumes unrelated to SRM are currently forcefully mounted to ESXi hosts on the recovery site, these LUNs are resignatured as part of a host rescan. So, all the virtual machines on these volumes become inaccessible and require re-registration. To prevent outages, ensure that no forcefully mounted snapshot LUNs are visible to ESXi hosts on the recovery site or set auto resignature mode to disabled. For more information, see VMware KB article 2010051.
- storageProvider.autoDetachLUNsWithDuplicateVolume: When multiple unmounted, unresolved snapshot volumes are found, SRM uses its API to detach these unwanted and unused snapshot LUNs. SRM then resignatures the volumes to be recovered and then reattaches the previously detached LUNs. This setting is only applicable when auto resignature is either set to disable or ignore. The default for this option is enabled.
This option is targeted for environments that have multiextent datastores with multiple copies of them presented to the recovery environment. Resignaturing in these situations can be complex and the functionality introduced by enabling this option eliminates the possibility of errors in regard to the resignaturing process. Dell's recommendation is to leave this option enabled. It is important to note that SRM only detaches detected snapshots of the devices to be recovered in the recovery plan.
Unmounted and unresolved devices, unrelated to the recovery plan (not in the recovery plan or snapshots of devices in it) will not be detached/re-attached. This option affects both recovery and test recovery operations.
The following figure shows the advanced settings options in SRM. Note that the settings are per site.