The table below lists and explains names and meanings of the DataKeeper parameters. These values are tuned by editing the /etc/default/LifeKeeper configuration file.
| Parameter Name | Meaning | Setting Value | Default Value | When to Apply | Notes |
|---|---|---|---|---|---|
| LKDR_CHUNK_SIZE | Sets the chunk size of bitmap in kilobits. | Integers | 256 | Creating a resource | This tunable can be set to any value from 0 to the size of the mirror. The value must be a power of 2. If it is not set to a power of 2 or is greater than the size of the mirror, the create of the mirror will fail. If it is set to zero it will default to 256k. If the value is too high, there’s not much benefit to the bitmap since huge chunks of data are being resynced for even one dirty block. If the value is too low, the bitmap overhead will be noticeable and will limit throughput on the mirror. |
| LKDR_CONNECT_NBD_DURING_RESTORE | Specifies if the NBD connection should be established when the mirror is restored (brought in-service). | True or False | True | To improve switchover and failover performance on a cluster with multiple mirrors and multiple targets | The default setting (“true”) provides the maximum data integrity as it assures that when a mirror is restored it will immediately replicate data to all targets available. To improve in-service performance (“false” setting) the NBD connections can be established during the first quickCheck interval (defined by LKCHECKINTERVAL, 120 seconds by default). While this will decrease the time required to bring the mirror in-service, it will also create an interval where data is not replicating to the targets. The performance improvement is most noticeable with multiple mirrors and multiple targets. |
| LKDR_SPEED_LIMIT | Specifies the maximum bandwidth that a resync will ever take – this should be set high enough to allow resyncs to go at the maximum speed possible. | Integers | 500000 | Resyncing a resource | |
| LKDR_SPEED_LIMIT_MIN | Specifies how fast the resync should be allowed to go when there is other I/O going on at the same time. As a rule of thumb, this should be set to half or less of the drive’s maximum write throughput in order to avoid starving out normal I/O activity when a resync occurs. | Integers | 20000 | Resyncing a resource | This setting must be less than the setting for LKDR_SPEED_LIMIT. |
| LKDR_ASYNC_LIMIT | Specifies the number of outstanding target writes that can be in flight at a given time. Increase this value for higher performance with asynchronous mirrors. | Integers | 4096 | Creating a resource | The valid range for this tunable is 0 – 16383. If it is set above 16383 the mirror create will fail. If it is set to 0 the async limit will default to 4096. If it is set to 1 it will default to 256. All other values between 2 – 16383 will result in an async limit set to that value. Note: Values below 4k are not recommended. |
| LKDR_NO_FULL_SYNC | Suppresses a full resync of newly added targets. | 0: not suppress 1: suppress |
0 | As required (takes effect immediately) | For detailed information, refer to “Avoiding Full Resynchronizations”. |
| LKDR_WAIT_TO_RESYNC | Specifies which parent resource(s) must be in-service before initializing mirror resynchronization. | False, hierarchy, <resource type> | filesys | Before resuming replication to verify replicated data is consistent. | False was the normal behavior in 9.5.1 and earlier. “Hierarchy” will wait until the full hierarchy is in-service. “<resource type>” like ‘filesys’ will wait for all parent resources of that type in the hierarchy to be in-service. |
| LKDR_WAIT_FOR_PREVIOUS_SOURCE_TIMEOUT | Specifies how long to wait for the previous source to join the cluster so that its bitmap can be merged. A full resync is required if replication is resumed to a target before the previous source’s bitmap is merged. This setting applies to all netraid devices; individual devices can NOT be configured with a different value. | 0: do not wait -1: wait indefinitely for the previous source to rejoin the cluster > 0 number of seconds to wait |
-1 | As required (takes effect immediately) | How SIOS DataKeeper Works |
Post your comment on this topic.