ATC Insight

You may be asking yourself from reading the title of this ATC Insight: "Why is there a Composer 2?" Has HPE Synergy really been around that long to go from Composer 1 to Composer 2? As a matter of fact it has been around for some time. HPE Synergy was released back in December of 2016. In this initial release, HPE Synergy was so ahead of its time that Composer 1 was released on Gen 8 hardware while the actual blades themselves were built on Gen 9 hardware. Composer 2 is built on the new Gen 10 platform and includes features like iLO 5 and Silicon root of trust. 

So why did HPE do the first HPE Synergy release on Gen 8 and Gen 9 mix? Well at the time that Gen 9 was coming out, HPE had already developed and tested Composer 1 on Gen 8 hardware, so they went with it. They needed to get HPE Synergy to market quickly in order to compete against Cisco UCS in what HPE coined as their "Composable Infrastructure."

Would you like access to an on-demand demo lab environment for HPE Synergy? Lab  

Synergy is HPE's answer to a multitude of things:

  1. It is a direct competitor to Cisco UCS and the Dell MX systems.
  2. It is the replacement for the aging C7000 chassis, which goes end of sale this year.
  3. When it is complimented with HPE OneView, Synergy allows you a much more stable upgrade path than the older C7000 standalone systems.
  4. It makes stacking C7000 Chassis much easier by linking the "frames" together with Composer and Frame link modules versus using the Onboard Administrator and Virtual Connect together.

Don't get me wrong, the C7000 was a really usable and stable platform for a long time, but it was a real headache during the upgrade process. I equate this to its inability to link everything with easy Profile Management.

So let's get into the meat and potatoes of the upgrade. We will only be upgrading the Composer from version 1 to version 2. We are not upgrading the Frame link modules. Why? It turns out we have streamers installed in our three frame solution and the 4 link frame modules are not supported with Image Streamer currently.

Below is an high-level design (HLD) of our current HPE Synergy environment:

  1. 2 x Synergy Composers v1
  2. 2 x Image Streamers
  3. 2 x Virtual Connect SE 40Gb F8 Module for Synergy
  4. 2 x Synergy 12Gb SAS Connection Module per Frame
  5. 3 x Synergy 12000 Frames
  6. Mix of Gen 9 and Gen 10 Compute Modules
  7. 3 x D3940 Storage Modules 1 per Frame
  8. 4 x Synergy 20Gb Interconnect Link Module
  9. 2 Port Frame link modules in each Frame
High-level design of WWT's HPE Synergy environment

Steps listed in below are high-level steps with a summary of what was observed from the upgrade. In the Documentation section of this ATC Insight, you will find detailed instructions from HPE in a PDF called "HPE Synergy Migration Guide." We followed this guide followed to perform the actual upgrade of our system. Also embedded below is a video that shows/explains the process in much greater detail, if you would like a deeper dive via step-by-step tutorial.

Pre-upgrade steps

All of these steps can be done by click on the OneView access pane on the left of your OneView Instance and go to settings.

1.) Always perform a backup of the environment before doing any type of upgrade that you are doing on this type of system.

2.) You must upgrade your current Synergy system to OneView 5.x. In our case, we upgraded our current 3 frame Synergy solution to HPE_Synergy_Composer_5.00.02.

3.) Record your network settings

Upgrade steps (high level)

Below is a break down of the upgrade steps to follow with some gotchas and things to look for along the way. Please refer to the video  below for more detail.

  1. Assuming you followed the Pre-Upgrade steps (steps 1-3 above) we are now ready to break the existing composer cluster. In doing so, we shutdown the standby composer and removed it from the cluster.
  2. Once the cluster was down to one node, we removed the current active node and installed the new Composer 2.
  3. We checked the current composer via the console of the Synergy frame and verified the current version of OneView matches the current version was just came from. We did this by logging in as the maintenance user and selecting details. If not, we would have first needed to re-image the composer that was shipped to the correct level. (This is done in two ways. Would could have used a usb stick or could have applied the network settings and upgrade once it was on the network)
  4. Once network settings were restored, we changed the Administrator password to your current Administrator password.
  5. Then, we were able to restore from backup. This process was fairly quick for us.  It only took 30 minutes and that includes the reboot of the composer.
  6. We logged in and verified that all the LIGs, Enclosure groups, and service profiles were in place.
  7. Verified alerts.
    1. The first alert you may run into discusses a cert issue that will need to be taken care of. To do this make sure to adjust your NTP settings on your appliance. The alert may not clear right away. It can take some time to clear. For us, it took about an hour then the alert just disappeared.
    2. Upload your SPP packs that you were previously using before the upgrade. As part of the process, HPE doesn't include it when creating the back up. It needs to be manually uploaded again after the upgrade.
  8. Next, we removed the old standby appliance and inserted the new Composer 2 standby appliance. Again, this one has to match firmware exactly. A caveat to this is if the firmware is different, then you will have to use the USB method to upgrade as the network upgrade option will not work.  The re-image process takes roughly 30 minutes (with the reboot process) to complete.
  9. After it is re-imaged successfully, the Composer 2 appliance will automatically connect to the cluster and start syncing with the database on the active node. It took about an hour for the sync to complete.
  10. Lastly, we verified alerts one last time to make sure there were no lingering alerts from the migration.

Impressions and final thoughts

Overall, the HPE Composer upgrade was fairly straight forward to follow in relation to the HPE PDF guide. We noticed a few documentation errors in the upgrade document we used (version used edition 3 published December 2019). But the document was easy to follow.

Overall, time of the upgrade from start to finish can vary drastically based on if you have to re-image your new Composers. If your Composer is already at the same version as your current HPE OneView version, then you are looking at roughly 3 to 5 hours. If not, you are looking at closer to 7 hours when you factor in re-imaging of the Composer nodes. I will note that while this is a lengthy process, one positive is you get off of the old Gen 8 server and onto a Gen 10 server. Another positive is that the infrastructure stayed up the entire time without any connectivity issues.

Expectations

Our expectations around the HPE Synergy Composer upgrade included the following:

  1. All infrastructure running in the Synergy domain continues to stay running with no downtime to operations on the Synergy compute modules that are installed and running.
  2. 3 to 5 hours needed for the actual upgrade process from start to end.
  3. Documentation from HPE is on point with step-by-step info needed for the upgrade.

Video tutorial

HPE Synergy Migration Guide (attached PDF): Detailed document from HPE that has step by step instructions on the upgrade and what we followed when doing this ATC Insight.

In the Video

Chris Braun is a WWT ATC lab veteran on the ATC Lab Services team. He is a deeply technical compute architect who is focused heavily into data center technologies around compute and hyper-converged infrastructure. He is also responsible for managing a team of ATC Lab architects who work with our customers on proofs of concepts (POCs) every day.

List of highlights from the Composer Upgrade video

  1. Backup - 0:46
  2. Verification of Synergy Environment - 1:45
  3. Network Settings - 1:37
  4. Remove Standby Node from Cluster - 2:14
  5. Shutdown Cluster - 3:40
  6. Connect to console of Composer - 4:00
  7. Set Network settings - 7:23
  8. Upgrade before restore of backup - 8:32
  9. Upload Backup - 11:34
  10. Restore Backup - 13:12
  11. Address and Identifier Removal - 16:47
  12. Install 2nd Composer and Re-image Process - 17:42
  13. Upgrade status and completion  - 22:31

Technologies