Solaris Storage Driver Known Issues of I/O Request Blocked for Disk Device
(Doc ID 1384697.1)
Last updated on OCTOBER 05, 2022
Applies to:
Solaris Operating System - Version 8.0 and laterSun Storage Traffic Manager Software - Version 4.6 and later
Sun Storage FC HBA - Version Not Applicable and later
Oracle Solaris on SPARC (64-bit)
Oracle Solaris on x86 (32-bit)
Oracle Solaris on x86-64 (64-bit)
Oracle Solaris on SPARC (32-bit)
Solaris 8 and later
sd/ssd driver
Mpxio
Failover operation
Symptoms
Under the following circumstances, i/o request can be stuck in the disk device driver without any message being displayed at the console. This can be seen with user command doing i/o or trying to access logical unit (disk) being block and not killable using kill(1M) or pkill(1M) command.
- Following a failover operation or any change in the configuration that will trigger a check of the path state of multipathed logical unit, read or write operation can be blocked.
- System resource starvation
This problem affects only configuration matching some of these criteria:
- Solaris version
- Solaris 10: MPxIO Multipathing module scsi_vhci version 1.73 or below
- Solaris 11: release 11.0.3.4.0 or below
- Solaris Multipathing (MPxIO) enabled to support active/passive or ALUA Storage array
- Failover operation happened following reboot of Storage array controller or loss of fiber channel link
NOTE. This is a generic document which deals with all known case of i/o being stuck, so not only when MPxIO is enabled, see below several and independent examples.
Changes
Most of the issues described in the document are seen after Failover operation triggered by reboot of Storage Array controller, loss of Fiber Channel link, etc ... or when the system is low on memory.
Cause
To view full details, sign in with your My Oracle Support account. |
|
Don't have a My Oracle Support account? Click to get started! |
In this Document
Symptoms |
Changes |
Cause |
1. I/O request are silently and endlessly retried following path failover or controller reboot/reset on asymmetric arrays |
2. I/O request are silently and endlessly retried due to DMA binding error |
3. I/O request are silently and endlessly retried due to incorrect logical unit state in Solaris kernel internal data structures |
4. I/O request are stuck in the sd/ssd driver and will never gets transported |
Solution |