C2 & ARM64 Instances end-of-life


Just received this email with above subject in our inbox. In summary :

As of December 1 st, 2020, our C2 and ARM64 Instances will reach their end-of-life. In addition, their support will end on July 1st, 2020.

This is preposterous !

We been with Scaleway for years since its inception and build lots of services over the years on these class of Servers. Since last 6 months, mysteriously we have been noticing these C2x (only) servers having intermittent loss of connectivity with the network storage; which necessitates a reboot/hard reboot. Multiple tickets to Scaleway technicians only coolly state they are aware of this issue - however fail to answer how come this issue was not occurring since the 2 years we were using these instances without any issues !

I believe Scaleway is not forthcoming on the real intent of sunsetting these servers - especially the dedicated C2x ones. The ARMs ones we have never have any issues since years!! Do you want us to again spent hundreds of hours rebuilding our architecture around those 3x times expensive new servers ( for similar dedicated servers specs ) before Dec 2020 - and hoping to be on our own since you have said no support on them past June 2020 - while we continuing paying you folks for them every month till your take it away !!!

This is totally unacceptable and unethical business practise, Scaleway !

I wish to know the feedback from the other clients on the same boat as us here.


They sent a survey few weeks ago, I said in the survey that I moved away most of our infrastructure from Scaleway because of stability and unavailability issues, I guess they received many negative feedbacks like mine and they decide to do something.
And this makes sense from a marketing point of view, if I weren’t forced to migrate my servers, i’d never try the new ones.

The only disappointing thing is the price of the new servers, 79 eur/month for a 8-core 32 gb RAM for example is much more than the competition (52 eur)


Will this affect START1-S and START1-M instances as well?
Do we need to migrate them or will they continue running.


I am very disappointed by this decision. I was using Scaleway because of the ARM offerings. After discontinuing ARM I will move to another provider I can trust more.
I didn’t have issues with ARM64 lately so it looks like they are improving the service in the background but my experience with customer service here is terrible. I see no reason why should I use Scaleway for x86 when there are plenty of better alternatives.
My opinion is Scaleway should improve their ARM offerings and use it to rebuild the trust their customer service lost.

1 Like

Same here. While I had stability issues with C2 instances and I had planned to move off from this offer due to this, this is not the case for the arm64 offer and the ARM* servers have been a key reason why I use scaleway.

It’s a shame that they put this to an end so abruptely.

1 Like

What’s more fishy is that Scaleway is replying to threads on this on HN & Reddit but not here!

Since, I’ve already said what I feel in the HN; I’ll use this space to sight a good example for handling EOL.

Heroku announced the depreciation of their Cedar14 stack on May
2018, EOL by April 2019 but they have extended it to Nov 2020 due to COVID-19. Although, it’s likely most have migrated by now.

I know that comparing SW stack of Heroku and HW stack of Scaleway is like comparing apples and oranges; yet it’s a good example for how a cloud provider should handle EOL irrespective of whether they are PaaS or IaaS, later should be more careful as the migration costs are higher especially with change in architecture (ARM64 -> x86).

P.S. I hope Scaleway doesn’t resort to manually degrading the performance of the said instance before EOL, sorry to say this but the way how this announcement is being handled has degraded the trust severely.

1 Like

Hello all,

Thank you for your feedback and sorry for not having answered here sooner.

As said on Reddit and Hacker News and as explained in our email, Scaleway’s C2 and ARM Instances suffer from random instabilities that prevent us to keep market them, even if they were not all impacted by these issues.

We would love to propose new ARM Instances in the future, but nothing is planned at the moment. We’re just investigating the hardware, but be assured that you will be kept informed.

Regarding START1 Instances and other former ranges, nothing is planned either, but we deeply recommend to migrate to our new Development and General Purpose Instances. They benefit from Block Storage and the latest features developed by our teams.


@tschmit I recently queried support to know if there was any deprecation plan for C1 servers because I’d like to be able to rely on their availability until 2022. The answer I got was this:

As of now, C1 removal has not been scheduled anytime soon and we do not have additional details to share regarding it’s End of Life status.
However, please note that we cannot guarantee that C1 servers will still be available on July 2022 in its current state.

So I’m going ahead with my project but I’m fearing that you will also deprecate them abruptly like you did for your ARM64 offering. I noticed they are no longer visible on the public website (or very well hidden) but they can still be ordered from the console…

For instance, I really dislike that you blocked creation of new instances on the day where you announced the deprecation. Some persons might have wanted some more instances to cover their needs until they had the time to handle migration to some other solution…

1 Like

No end of life is indeed planned for the moment for C1 Instances. Nevertheless, these instances are now almost six years old, which explains why they are no longer presented on our site, and we cannot guarantee that they will still be supported in 2022. We recommend to rather turn to the Development and General Purpose Instances to build a future-proof infrastructure.

Concerning C2s and ARMs, the creation of new instances is no longer possible as of today for customers who do not use these offers, but we have considered those who still use them. Their quota per offer has not been reduced to zero, but to the number of instances they have plus one, which lets some flexibility to prepare migration.

You can be sure that we are doing everything we can to ensure this end of life is not abrupt, but rather smooth. That’s why we are proceeding step by step before December 1st, 2020.


There is no real migration path from ARM64 to DEV or Baremetal. We read your migration document. The rsync-path is just to copy the data binary-to-binary from one volume to another. We will have to reinstall the new OS, system and application software, configure and then apply the updates !
For most of us, with years of association with Scaleway who used ARM64 nodes these is a gargantuan amount of work !

We need a real migration from ARM64 to another ARM64 alternatives. WE do not believe these current ARM64 instances has any issues as we did personally not face any in last 2+ years we uset hem. This EOL exercise, we believe is done to move clients from this very efficient, low carbon footprint and low-cost servers to expensive virtual and bare-metal instances Scaleway is touting now !!

You are being warned against this ill-conceived marketing ploy and annoying your loyal client base . There are now real alternatives out there …

1 Like

Instead of making local boot work on arm64 work, as “promised” in the help articles, you just decided to throw them out!?

Congratulations, you are now a walking advertisement for Amazon Web Services.

I hope you’ll at least sell the ThunderX hardware on the used market instead of turning it into e-waste.


1 Like

I’d be interested to know what are the alternatives you are referring to @thomasch. I have found that hosted ARM* hardware is not very common to say the least and I was pleased with the scaleway offering so it’s a pita that they are shutting it down.

1 Like

will PM you as not appropiate to discuss this here.

1 Like

Another details about the C2X stability, we have noticed since last 8 months ( everything was fine with these C2x instances for the first 2 years we used them )… these Servers will loose connectivity with NAS in 8 out of 10 times on a Saturday early morning GMT. The only recourse then is to reboot them from console. We are curious what kind of intervention is done at the SCW end which causes this issues ? They do not answer this …

1 Like

At a guess, that’s when they do backups. They’re notoriously resource-intensive. I can always tell when the nightly backups are happening on LeaseWeb VMs because I/O perf goes through the floor for two hours. Unfortunately, that’s our two busiest hours every night. (Update: They seem to have fixed this as of May 2020.)

1 Like

Scaleway, important notifications like end of life of 2 services should be on your blog. If you dare to discontinue the service with such a short notice don’t be shy to let everyone know about it.

Anyway, the good thing about Scaleway its’ mediocre support made me able to restore the service I ran here in a few hours using any other provider.

1 Like