Two year old bug still not fixed
Today I’ve updated a Datacore-Cluster from SanSymphony10 PSP6 U1 to PSP6 U4. In the past we had issues with updating DataCore-Emulex drivers during the update process. Some updates went fine – others resulted in a BSOD. That was in 2015. Customer Support told us, that this was fixed with SanSymphony10 PSP1.
As a security precaution I’ve followed a procedure (article in german) where I first uninstalled all Datacore-Emulex drivers before I applied the DataCore update and replaced drivers on all HBA afterwards. That is time-consuming and complicated. So I gave it a try and did a regular update to a DataCore cluster with Emulex HBA (12000 series).
Bad idea
On the first host the update went fine, but on the second it ended up with a BSOD and a reboot. Luckily the system wasn’t damaged by the crash this time. Believe me, in the past I had a damaged system too – no fun.
Steps to recover
During crash your server might declare one or more HBA as damaged (but they are ok). You will have to enable them again in the BIOS und boot your OS normally.
Make sure DCSX is not running (it shouldn’t because you were instructed to set it to manual startup before the Datacore update).
Open a CMD prompt and change directory to:
cd C:\Program Files\DataCore\SANsymphony\Install
Execute the next command to uninstall your current Datacore-Emulex HBA drivers:
DcsDriverSetup /uninstall:DcsFcSli.inf /verbose
Reboot your DataCore Host and launch the update setup again (it will resume – no worries).
When the update is finished, you’ll be asked to reboot your host. Choose a later reboot and change HBA drivers first. Go to server manager > Storage Controllers > Emulex LightPulse HBA. Replace driver and browse locally for drivers. Point the wizard to the local path:
C:\Program Files\DataCore\SANsymphony\Install
Repeat for all HBA. Reboot host after all drivers have been replaced.
Why does this still happen?
According to customer support this should have been fixed in PSP1 – now we have PSP6 Update4. More than 18 months have passed since I’ve seen that issue for the first time. Still not solved.
My advice for the future: When updating DataCore Clusters with 12000 series Emulex HBA, follow my update-procedure that I’ve posted in 2015. That posting ist written in german, but with the information above and the screenshots you should be fine. Do not hesitate to ask me if some instructions are not clear.