ESXi 7.0 U3c - your experience?

[PatrickLong]

Looking for any comment from admins out there who have already upgraded their Gen9 / Gen10 HPE servers to ESXi 7.0 U3c - especially with diskless servers (i.e. boot from microSD).  I have a LOT of diskless servers BL460c Gen 9 and SY480/660 Gen10's running 7.0 U2d which has been rock solid imo, but in order to get the latest security patches we'll need to go to either 7.0 U2e or 7.0 U3c.  The early misadventures with early 7.03 line of releases give me a bit of pause; just looking for an comments from someone who might have already upgraded using the HPE Custom image profiles for ProLiant: VMware-ESXi-7.0.3-19193900-HPE-703.0.0.10.8.1.3-Jan2022

and for Synergy:  VMware-ESXi-7.0.3-19193900-HPE-703.0.0.10.8.6.3-Jan2022-Synergy

Have you seen any issues with 7.0 U3c post-update?

Yes, I know that exclusively SD (or in fact ANY USB-based) boot devices/diskless servers is no longer a supported boot method beginiing with future major releases of ESXi (8.x), but it is still fully supported per KB85685 and it will take me a few months to retrofit the boot devices of my existing hosts - I still have to address security issues in the interim.  Thoughts?

[support_s]

System recommended content:

1. HPE VMware ESXi 7.0 U3 Customized Image Release Notes for January 2022

2. HPE VMware ESXi 7.0 U2 Customized Image Release Notes for May 2021

Please click on "Thumbs Up/Kudo" icon to give a "Kudo".

Thank you for being a HPE valuable community member.

[PatrickLong]

TFW you get all excited about a new response notifcation to your Community post....only to discover it's the support_s bot account tangentially matching an OP text string.  

[PatrickLong]

Answering my own question - I upgraded one of my SY480 Gen10 compute modules using the Synergy=specific image profile - no issues.; however, the same cannot be said for the single BL460c Gen9 that I upgraded.  This server uses the 630FLB CNA adapter (HP FlexFabric 20GB) based on Qlogic 57840 10/20 adapter chipset at the current firmware and driver level.
firmware: HP FF 20GB 2-port 630FLB BCM NetXtreme II BCM57840 firmware 7.18.82 iLo / MFW 7.16.3 / FW 7.13.18.1
network driver: qfle3 1.4.17.0-1OEM.700.1.0.15843807   /    fc driver: qfle3f 2.1.25.0-1OEM.700.1.0.15843807 

and after the 7.0 U3c upgrade via ProLiant image profile as shown:

esxcli software profile install -d /vmfs/volumes/<datastore_name>/<iso_directory>VMware-ESXi-7.0.3-19193900-HPE-703.0.0.10.8.1.3-Jan2022-depot.zip -p HPE-Custom-AddOn_703.0.0.10.8.1-3 --ok-to-remove --no-hardware-warning

1st reboot - server came up and did not reconnect to vCenter despite DCUI fully up and accessible showing no issues; host did not respond to ping at all.
local ESXi shell Alt-F1 showed only 2 physical NICs present using command esxcli network nic list (should show all 8 NIC multi-functions) but cdid not pass any of the network troubleshooting tasks, couldn't access local gateway IP, DNS, etc.,

2nd reboot - this time it came up respoinding to ping and the two mgmt NICs were active, but again, only showed two NICS via esxcli network nic list command instead of all 8 Multifunction adapters.  This is not normal.

3rd reboot - this time it came up with all 8 NIC functions intact and functional as follows:

4th reboot - same - all 8 present

5th reboot - same - all 8 present

6th reboot, etc.  – same – all 8 NIC functions present...

WHAT could account for this inconsistent enablement of different numbers of NIC functions on the first few reboots after the 7.0 U3 update was applied?  Anyone have any ideas?

Going to try a few others to see if this behavior repeats itself.  Will post findings here.

<Edit 2/24/2022>  I have updated 11 more identically-configured BL460c blades using the same method described above, and did not experience this issue again; all 11 successfully upgraded..  Unfortunate timing for the one upgrade issue/failure to be the very first BL460c I upgraded..

[SwerveShot]

In light of the nearing EOS date for vSphere 6.7 one of my clients upgraded to vSphere 7 U3f at the beginning of August. We did our due diligence and checked the VMware HCL and all relevant HPE resources. We made sure we were on the latest supported SPPs and used the HPE provided ESXi ISO. All should be well.

However after upgrading especially on the ProLiant BL460c Gen10 blade servers we are experiencing all sorts of issues. The main issues are frequent PSODs and disconnects from the network (using HPE virtual connect). These issues can be resolved by rebooting and/or performing a E-Fuse reset. But in an environment consisting of about 90 ProLiant BL460c Gen10 blade servers we experience about 2 failures per week. Ouch!

We don't have any active service contracts for these machines so we are looking towards the online communities to see if anyone has similar issues. Without any luck thusfar. These servers are of course at the end of their lifecycle but we hoped we could use them just a bit longer before replacing them. A rollback to vSphere 6.7 sounds more appealing with each day.

---

UPDATE 11/21/2022
Instead of rolling back to vSphere 6.7 we decided to perform a clean install of vSphere 7. It's been almost a month without the dreaded PSODs. Sadly, the random network link failure issue still remains.