SimpliVity Battery error, SSD cache disabled issues

[BoonHong]

Batteries and harddisks have an average lifespan of 3-5 years. And when our SimpliVity cache batteries started to fail one by one, I'm dismay to find out that HPE support told us that failing component such as the SimpliVity cache battery, will not be replaced until they actually failed. This to me is reactive maintenance, not proactive. (FYI,  Dell has previously replaced all our SAN storage drives after 5 years when we start to have regular drive failure.)

 In this case, can we purchase spare battery module to replace ourselves after these nodes warranty has expired? Since Gen9 warranty is expiring and there is no option to extend them anymore?

I’m also disappointed that HPE support took 5 days just to get the battery replaced. For Dell, they are able to do it within 4 hours. Support Case: 5364549297

In additional, after the battery replacement, the OVC became faulty. And HPE support need a new support case to deal with this: Support Case: 5364688532

The OVC issue was resolved by upgrading it to version 4.1.1u1, and HPE support told us that the other nodes need to quickly upgrade to the same version as well.

However, the battery error and cache disabled resurfaced on the same node again despite having battery replaced.

Is it safe to shutdown another node to upgrade the OVC in a 3 nodes cluster where one already have cache disabled?

For extra safety, can we temporary remove one node so that the federation is relying on 2 instead of 3 nodes? This means storage consumption will increased from 66.6% to 100% per node right? How do we ensure that the existing free space is sufficient for 2 copies of all data to be compact into 2 nodes from 3?

[support_s]

System recommended content:

1. HPE SIMPLIVITY

2. SimpliVity - Battery Backup Health Alarm Triggered on Reboot on Versions 3.7.9 & 3.7.10

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

Thank you for being a HPE valuable community member.

[gustenar]

Hello @BoonHong, 

Regarding the parts, if they are available you can purchase them if you'd like. You may want to talk to an authorized HPE reseller for availability or our parts page. 

To address your other question, if one node is currently shutdown and you turn off a second one, the VMs that share replicas between those two nodes will be affected and lose access to storage. If you need to upgrade, it's better if you shutdown the VMs in that case. 

If you remove a node from federation the replicas that lived in that host will try to be recreated in the remaining nodes of the federation, the actual storage consumption gain will be based on the sizes of those VMs. You need to make sure you have the necessary space available in the remaining nodes for those replicas to rebuild properly. 

[BoonHong]

Hi Gustenar,

Thanks for your helpful replies.

1. HPE told us that they do NOT sell parts as per HPE policies. Can you verify if this is actually not true?

>> Unfortunately, you cannot order a spare battery module for future as per HPE Policies

2. Most of our VMs are required 24/7. So we are unable to shut them down. HPE support suggest that we upgrade our firmware 4.0.1 to supported version v4.1.0 first, but we do not want to risk rebooting 1 for the upgrade, while having another unstable node in the federation that has battery error and cache disabled.

3, Is there a way to calculate or find out the actual storage consumption gain if we remove a node? With a node removed, we will be able to have another node reboot without losing access to storage right? This help address the concern and issue at point 2.

[gustenar]

As mentioned before, for parts contact an HPE reseller for assistance. I believe what has been indicated by support group is that they can't send parts that are not defective. Of course, if your server still has a hardware issue it can be addressed by the support team if the host is still under support/contract. 

I think the best approach is to fix the hardware issues on that host so you can have an all healthy federation and then plan on the software and firmware upgrades without the need of removing any hosts. 

[BoonHong]

@gustenar 

The error is not resolved after the faulty hardware is replaced. The engineer feedback that the replacement part look completely difference despite the part number is correct.

The support team suspect the firmware of the replacement part is not compatible with our current version and suggest that we upgrade our SimpliVity from v4.0.1 to v4.1.0. This become a chicken and egg issue as we do not want to take the risk of shutting down a node for the upgrade while we have another node having SSD cache disabled issue.

In additional, there is a storage network error. Support team claims that we must change our storage NICs from Active-Active to Active-Passive. But all our nodes storage NICs have always been in Active-Active mode without any issue, as done by the HPE deployment team. Are we to amend all to Active-Passive now?

[gustenar]

It may not be possible to upgrade firmware on a defective part, please reach out to the support group for assistance with the part. Regarding your network question visit this link:

https://www.hpe.com/psnow/doc/a50000783enw

Review the NIC teaming requirements under page 9.