The following table lists the issues that are fixed in PowerFlex3.6.0.5.
Issue number & SR number | Problem summary |
---|---|
SCI-65334
SR# 134810840 |
During non-disruptive upgrade to PowerFlex 3.6, the SDS maintenance mode takes longer when the MDM returns an incomplete query to the PowerFlex Gateway. |
SCI-68154 | The SDC container used supervisord to run and log the diagnostic message. All the kernels do not support the diagnostic message switch which caused the supervisord to return the diagnostic messages in an infinite loop and consume excessive CPU. |
SCI-67542
SR# v1007162 |
Disruptive upgrade occurred on SDC during RCM upgrade. Deadlock on locking Octet in L2P directory between the L2P Page merges flow. The L2P split-upgrade flow that is invoked causes a disruptive upgrade due to SDS IO timing out. |
SCI-68153 | The SDC-SDS TCP disconnected when the SDS host was hard reset while establishing the security communication link. The keepalive manager does not close the socket due to inactivity and remained open indefinitely. |
SCI-65052
SR# 133347603 |
The SDS disconnected when a large set of data was allowed into the Protected Maintenance Mode. The MDM delayed the RoleSwitch and created a timeframe which resulted in an IO error until the SDS responded to all the combs. |
SCI-66526
SR# 139180450 |
During an IO background operation, you must split few operations to multiple MD pages to fit in the correct size of an FG device. A new write with an existing background operation leads to an overflow in the MD buffer space and causes the SDS crash when detected. |
SCI-66575 | IM scripting “stop process on script failure" does not stop on failure and the indicator may be overridden. If validateNoCriticalFailureCommand indicates a failure due to a flag in criticalCommandFailedArr, the sequence stops at exit-SDS-from-MM since this command always runs. |
SCI-65531 (SR# 135236911) | When you unmap a volume from the latest SDC version, the volumes remain mapped even if the unmap operation is completed. |
SCI-67424
SR# v1005052 |
SDS decoupled due to panic "ALWAYS_ASSERT IO takes too long". Incorrect locking order of L2P Directory octet and octet counters leads to a deadlock between L2P Page merge flow and QueryCapacityUtilization. This blocks many IO operations indefinitely until SDS panics due to timeout. |
SCI-67007 | Execution of sas3ircu triggers SVM crash on PowerEdge 15G-based hosts. For details, see DTA 199549. |
SCI-66906 | Tomcat versions 8.5.79 and earlier is exposed to CVE-2022-29885. PowerFlex Gateway does not use the clustering feature and is not exposed to the issue. |
SCI-65370 | The exclude log collection check box for the SDC log collection was removed. The check box is restored to exclude and include the node log collection. |
SCI-62416
SR# 119473468 |
When running scripts on host feature, the operation logs does not check the script files in nodes. |