Timeout Values
Timeout Values for iSCSI and FC Arrays
Various timeout values affect Nimble targets from Windows hosts with and without Multipath-IO (MPIO). Refer to these recommended values for various configurations. Note that the LinkDownTime and MaxRequestHoldTime values apply to iSCSI arrays only. NWT sets all these values, which work with arrays running NimbleOS 1.x or later.
| Registry Key | Meaning of Value | Default Values (in seconds) |
|
|---|---|---|---|
| Windows | Nimble | ||
| HKEY_LOCAL_MACHINE\SYSTEM \CurrentControlSet\service\Disk\TimeOutValue | Global disk timeout value for disk.sys. If timeout value is met, OS sends bus reset to all LUNs. | 60 or 120 (OS dependent) |
60 |
| HKEY_LOCAL_MACHINE\SYSTEM
\CurrentControlSet\Control\Class\
{4D36E97B-E325-11CE-BFC1-08002BE10318} \0001
\Parameters\MaxRequestHoldTime (See Note) |
(iSCSI only) Maximum time (in seconds) for which requests will be queued if connection to the target is lost and the connection is being retried. (Applies only to non-MPIO disks.) |
60 | 120 |
| HKEY_LOCAL_MACHINE\SYSTEM
\CurrentControlSet\Control\Class\
{4D36E97B-E325-11CE-BFC1-08002BE10318} \0001\Parameters\LinkDownTime
(See Note) |
(iSCSI only) Determines how long (in seconds) I/O requests will be held in the device queue for a MPIO path and retried if the connection to the target is lost. |
15 | 30 |
| HKEY_LOCAL_MACHINE\SYSTEM \CurrentControlSet\services\mpio\Parameters \PDORemovePeriod | Time (in seconds) that the multipath pseudo-LUN (Physical Device Object in the kernel) will continue to remain in system memory, even after losing all paths to the Nimble volume. | 20 | 90 |
| HKEY_LOCAL_MACHINE\SYSTEM \CurrentControlSet\services\mpio\Parameters \UseCustomPathRecoveryInterval | UseCustomPathRecoveryInterval (Boolean 1 or 0) and PathRecoveryInterval times (in seconds) can be used to perform path recovery while the PDORemovalPeriod timer counts down. | 0 | 1 |
| HKEY_LOCAL_MACHINE\SYSTEM \CurrentControlSet\services\mpio\Parameters \PathRecoveryInterval | 40 | 40 | |
| HKEY_LOCAL_MACHINE\SYSTEM \CurrentControlSet\services\mpio\Parameters \RetryCount | Number of times a failed I/O if DSM determines that a failing request must be retried on current path to a LUN on certain transient errors. | 3 | 30 |
| HKEY_LOCAL_MACHINE\SYSTEM \CurrentControlSet\services\mpio\Parameters \RetryInterval | Interval of time (in seconds) after which a failed request is retried (determined by DSM, and assumes that the I/O has been retried a fewer number of times than RetryCount). | 1 | 1 |
| HKEY_LOCAL_MACHINE\SYSTEM \CurrentControlSet\services\msdsm\Parameters \DsmMaximumStateTransitionTime | Time SCSI Reservation is retried on a failed path for specific CHECK_CONDITIONS raised by the array. | 3 | 120 |
- All values are in seconds.
- A reboot is required after value changes.
- The iSCSI initiator driver instance ID (for iSCSI only values) may be different than 0001.
FC HBA Timeout Values
The following values apply when configuring timeouts on FC HBA ports, depending on the type. The Emulex management tool OneCommand Manager allows you to configure timeouts on a per-port basis. Similarly, Qlogic's QconvergeConsole allows per-port configuration under . Note that reboots are required for Qlogic FC HBAs.
| Timeout Type | Value | Meaning of Value | Default Value (in seconds) |
Range (in seconds) |
|---|---|---|---|---|
| Emulex FC HBA | LinkTimeout |
Time interval after which a link that is down stops issuing a BUSY status for requests and starts issuing SELECTION_TIMEOUT error status. This includes port login and discovery time. (No reboot required.) |
30 | 0-255 |
| NodeTimeout |
Time interval after which a prior logged-in node issues SELECTION_TIMEOUT error status to an I/O request. This causes the system to wait for a node that might re-enter the configuration soon before reporting a failure. The timer starts after port discovery is completed and the node is no longer present. (No reboot required.) |
30* | 0-255 | |
| Qlogic FC HBA | LinkDownTimeout |
Timeout value when a link that is down stops issuing a BUSY status for requests and starts issuing SELECTION_TIMEOUT error status. This LinkDownTimeOut includes port login and discovery time (Reboot required.) |
30 | 0-255 |
| PortDownRetryCount |
Number of times the I/O request is re-sent to a port that is not responding in one second intervals. Alternately, this setting specifies the number of seconds the software waits to retry a command to a port returning port down status. (Reboot required.) |
30 | 0-255 |