Article Number: 000188498
In a version mismatch situation where a VPLEX firmware upgrade fails and rolls back, we may hit a situation where the VPLEX VS6 MMCS-A (management server) is on the higher firmware version (6.2.x) and the directors are on the original lower firmware version (i.e. 6.1.x). This is causing a cosmetic issue with the VPLEX user interface (UI) resulting in all storage volumes in the system reporting the 'operational' and 'health' status as 'unknown'. However, on closer inspection the volumes are alive and the top-level virtual volume is not degraded.
1] Confirm there is a code mismatch. This can be done using the VPlexcli “health-check” command and/or the VPlexcli “version-a” command.
The example below is from the top section of the health-check output:
Product Version: Version mismatch (or NDU) << mismatch indicates there is a different firmware version between the management server and directors Product Type: Metro WAN Connectivity Type: FC Hardware Type: VPL <-- represents the VS6 Cluster Size: 4 engines <-- says this is Quad Engine configuration, 2 = Dual and 1 = Single Cluster TLA: cluster-1: CKMXXXXXXXXXXX cluster-2: CKMXXXXXXXXXXX The storage-volume issue is most apparent in the VPlexcli command “storage-volume summary” output and is also visible from the Back-End (BE) Storage portion of the 'health-check' command output. The issue is not reported in the 'ndu pre-check', the 'connectivity validate-be', nor the 'cluster-status' command outputs. Example from VPlexcli outputs provided below. Before the failed NDU attempt no storage volumes will report as 'unknown', yet after the failed NDU we will see the following, the storage-volume 'IO status' will equal 'alive', however, the 'Operational Status' and the 'Health State' will equal 'unknown' for all storage volumes. VPlexcli:/> storage-volume summary SUMMARY (cluster-1) StorageVolume Name IO Status Operational Status Health State --------------------------------------- --------- ------------------ ------------ VCKM001530XXX1-00003 alive unknown unknown << observe VCKM001530XXX2-00004 alive unknown unknown << VCKM001530XXX3-00006 alive unknown unknown << . . Storage-Volume Summary (no tier) ---------------------- --------------------- Health out-of-date 0 storage-volumes 4372 << note the total number of storage volumes in the system unhealthy 4372 << note the total number of volumes equal unhealthy, all storage volumes in the system are now reporting unhealthy. Vendor DGC 1276 XtremIO 3096 Use claimed 1 meta-data 4 unclaimed 5 used 4362 Capacity total 2.92P From the VPLEX 'health-check' command output scroll to the section called “BE Storage” and check the “Unhealthy Storage Volumes” to see all volumes are reported Unhealthy. BE Storage: << ----------- Cluster Total Unhealthy Total Storage No Not visible With Total Name Storage Storage Provisioned/ Dual from Unsupported Extents/ Volumes/ Volumes Limit Paths All Dirs # of Paths Limit Limit --------- -------- --------- ------------- ----- ----------- ----------- -------- cluster-1 4372/12000 4372 2.92P/8PB 0 0 0 4362/24000 cluster-2 4372/12000 4372 2.92P/8PB 0 0 0 4362/24000
FE Storage: ----------- Cluster Total Unhealthy Total Dist Unhealthy Local With unsupported Name Virtual Virtual Devs/ Dist Top-Level RAID1 mirror Volumes/ Volumes Limit Devs Devices/Limit legs Limit --------- ---------- --------- ---------- --------- ------------- ---------------- cluster-1 3794/12000 0 1024/12000 0 2770/12000 0 cluster-2 4356/12000 0 1024/12000 0 3333/12000 0
Generally, it is okay to operate with the firmware mismatch for a period of time however, it is not recommended to remain in this state for too long and to resolve this issue as soon as possible.
Resolution:
This issue is resolved once the director firmware upgrade is completed, bringing the directors and management server onto the same matching code version. This is the preferred option to resolve this issue and this option should be applied.
As stated earlier in this article, this is a cosmetic issue and it is okay to proceed with the director firmware upgrade while this issue is present.
This issue, although cosmetic, can cause other related issues such as:
VPLEX, VPLEX GeoSynchrony, VPLEX Series
29 Aug 2022
6
Solution