#Questions: Cluster Switch Update (BES-53248-A2)

1 messages · Page 1 of 1 (latest)

valid bear
#

Hi everyone,

Currently, I'm planning to perform cluster switch update on BES-53248-A2.
FYI,
AFF A300 ONTAP-9.15P16
FAS8200 ONTAP-9.15P16
BES-53248-A2 EFOS-3.12.0.1 RCF-1.12

Based on my findings on NetApp Hardware Universe,
EFOS 3.14.0.2 is supported for 9.15.1P16, (correct me if I'm wrong)
so we proceed the update from EFOS 3.12.0.1 to EFOS 3.14.0.2.
SUMMARY ON [UPDATE EFOS] documentation,
https://docs.netapp.com/us-en/ontap-systems-switches/switch-bes-53248/upgrade-efos-software.html#prepare-for-upgrade

  • verify non-FIPS to non-FIPS
  • proceed method 1
  • ensure .stk is prepared
  • disable auto-revert on cluster LIFs
  • download .stk to backup
  • boot system from backup
  • wr and reboot
  • repeat steps on switch 2

QUESTIONS

  1. Once the EFOS update is completed, can we set the ‘backup’ image to EFOS version 3.12.0.1? My intention is to ensure that if the primary active boot image fails, we still have a backup image with an older version
  2. Based on the documentation, EFOS-3.14.0.2 is only supported with RCF v1.13. Will there be any compatibility issues if we remain on the older RCF v1.10? Or do we need to update the RCF as soon as possible to avoid triggering errors or causing the cluster network to go down?

SUMMARY ON [UPDATE RCF] documentation,
https://docs.netapp.com/us-en/ontap-systems-switches/switch-bes-53248/upgrade-rcf.html

  • verify cluster health and port status
  • disable auto-revert on cluster LIFs
  • shutdown cluster ports
  • clean config (paste given config to remain remote connectivity)
  • wr and reboot
  • apply new RCF config to "reference-config.scr"
  • wr and reboot
  • bringup cluster ports
  • wr and verify status
  • repeat steps on switch 2

QUESTIONS

  1. Just in case the given config does not work, we can still use the console to perform the basic setup, right?

That concludes the steps.
Would appreciate it if could point out any concerns or steps to take note of avoiding any cluster network failure.

Thank you!

serene zealot
#

Just a note
After you turn off the auto revert on the cluster lifs they will only move if they have to.
You mention in the summary to bring up cluster ports. Doesn’t happen that way

Auto revert is off. You shut down ports on one switch which forces all lifs to one switch. They stay there until you shut the ports down on the second switch. This is why it’s important to verify (verify ports are up and heathy and not degraded before continuing).

So after you think the first switch is good, you go to the second switch and run the process again

At the end the ports are still on one switch.

You’re advised at the end to enable auto-revert on the cluster lifs. They still will not go back home until ONTAP detects a link change. This is why the instructions ask you to reboot the last switch one more time

valid bear
valid bear
serene zealot
#

Not necessary
Upgrade to 3.12+ and use the reference-config.scr

It will automatically detect all licensed ports and activate only those

#

The reference config will only work when the EFOS is 3.12+

valid bear
#

understood, jst worry it will trigger any errors when pushing unlicense port config

#

Lastly, how long basically does the upgrade process take?

serene zealot
#

If you use a standard file, it absolutely will. If you use the reference config it does not

I personally would have a pair of switches done in less than an hour. Point of information: I do them frequently(both Cisco and BES)

#

They much faster than Cisco (3-5 minutes, Cisco can take 7-10 to be ready)

#

If I’m doing it, I skip a bit

Disable auto revert
Shut down switch ports
Erase appropriate commands per script
Save config
Upgrade EFOS
(Requires reboot)
Install updated rcf.
Reboot again
Done. Repeat

#

If you have not done the process, please follow the will documented process

I helped correct it a few years ago

valid bear
#

yup, currently will jst follow step by step from given documentstion

#

That helps alot anyway, appreciate yr responds, thank you

valid bear
#

when we apply new RCF on reference-config, we dont need to include mgmt config like hostname, serviceport ip, domain name, ntp etc into new RCF right?

#

bcs from documentation, we only clear config on interface ports and service policy related info, those existing mgmt info will be remained even new rcf is applied

#

this action can remain remote connectivity as well

serene zealot
#

If you follow directions exactly, the switches will stay available remotely

If you mess up the EFOS update then you will need a tftp/http server to reinit the switch. I inadvertently skipped a step one time and the boot got borked

#

Following the directions will only clear the rcf config and leave the service port and and any other customizations alone

valid bear
#

alright noted on this, thank you!