1) i just did this upgrade on 4 clusters. It's a normal rolling reboot, one node at a time. I don't ever reboot my clusters prior to the upgrade.
2) I always upgrade DR first and Prod the next day
3) Not sure what you mean, same as before where CIFS clients get disconnected and NFS clients get moved to other nodes
4) I can't think of customer configuration that would not survive an upgrade, my custom sysctl settings seem to survive between upgrades. For example a sysctl settings for session TTL.
just like to agree with Dynamox, did a 20 node last weekend, other than the complete guesstimate of time for the upgrade ( out by 40 % ). Upgrade was seemless to end users and no config was lost.
Few followup uestions, appreciate if you guys could help.
1) I have seen customers doing DR upgrade first, is this because it is a safe approach and does rebooting the DR cluster has same result , i,e SMB users will get disconnected ?
2)Also, We are going for a major upgrade , ie. 8.0 to 8.1 and the document says it needs to be a simultaneous upgrade . Can we still do a rolling upgrade ?
3)finally , CE is asking to better power cycle the cluster as it has not been done since one year ..In case we have to do a powercycle should I inform the SMB clients about the activity ? It will be done during off hours(weekend if possible) and I am not expecting users getting impacted ,but wanted to know if SMB and NFS users need to be notified for the upgrades ?
From where can we download the tool -EMC Assessment Workbench, which has a sub-component , File System Assessment included in this tool (FSA), exceptionally useful when trying to determine workloads, file types, file sizes, access trends, growth, etc.
dynamox
9 Legend
•
20.4K Posts
1
August 16th, 2018 04:00
1) i just did this upgrade on 4 clusters. It's a normal rolling reboot, one node at a time. I don't ever reboot my clusters prior to the upgrade.
2) I always upgrade DR first and Prod the next day
3) Not sure what you mean, same as before where CIFS clients get disconnected and NFS clients get moved to other nodes
4) I can't think of customer configuration that would not survive an upgrade, my custom sysctl settings seem to survive between upgrades. For example a sysctl settings for session TTL.
cadencep45
3 Apprentice
•
318 Posts
1
August 16th, 2018 04:00
just like to agree with Dynamox, did a 20 node last weekend, other than the complete guesstimate of time for the upgrade ( out by 40 % ). Upgrade was seemless to end users and no config was lost.
hammerfall25
2 Posts
0
August 21st, 2018 13:00
Thank you Dynamox and Castleknock .
Few followup uestions, appreciate if you guys could help.
1) I have seen customers doing DR upgrade first, is this because it is a safe approach and does rebooting the DR cluster has same result , i,e SMB users will get disconnected ?
2)Also, We are going for a major upgrade , ie. 8.0 to 8.1 and the document says it needs to be a simultaneous upgrade . Can we still do a rolling upgrade ?
3)finally , CE is asking to better power cycle the cluster as it has not been done since one year ..In case we have to do a powercycle should I inform the SMB clients about the activity ? It will be done during off hours(weekend if possible) and I am not expecting users getting impacted ,but wanted to know if SMB and NFS users need to be notified for the upgrades ?
TIA
cadencep45
3 Apprentice
•
318 Posts
0
August 22nd, 2018 08:00
answers in a random order..
3
Have not had box long enough to answer..
2.
Suggest you open SR with EMC and get answer that way
1.
Do not have config to say. Suggest you open SR with EMC and get answer that way
Bhuppi1
2 Intern
•
131 Posts
0
March 22nd, 2019 14:00
Hi Dynamox,
From where can we download the tool -EMC Assessment Workbench, which has a sub-component , File System Assessment included in this tool (FSA), exceptionally useful when trying to determine workloads, file types, file sizes, access trends, growth, etc.
Thank you.