Ahh yes. I had started writing a ticket for this but never submitted it. This is a big +1 from me. I can say that in the past year that we've been generating AMIs, not a single person has ever asked me (nor anyone I know of) for a 32 bit image. In fact, only like 4 tiny instance types on AWS even support it.

I've talked about this with a number of people in #fedora-cloud who seem to be on board with ceasing production of the 32 bit images. We could probably bring this up in the meeting and on the list, so that we can hear from anyone who disagrees with our assessment.

Replying to [comment:1 oddshocks]:

I've talked about this with a number of people in #fedora-cloud who seem to be on board with ceasing production of the 32 bit images. We could probably bring this up in the meeting and on the list, so that we can hear from anyone who disagrees with our assessment.

Can you send an email to the list where we can +1 or -1? We can also discuss in meeting next week, but having a vote will help make this start to become real.

Unless someone can come forward with a compelling "we desperately need 32-bit for X" use case, then let's streamline and get rid of the 32-bit images.

+1

+1

+1 from me.

I think that without having download/user statistics we can't really measure it well. The only way to tell if it's being used is to pull it, and be ready to put it back if there's enough noise about it.

+1 to getting rid of it.

Replying to [comment:6 roshi]:

I think that without having download/user statistics we can't really measure it well. The only way to tell if it's being used is to pull it, and be ready to put it back if there's enough noise about it.

+1 to getting rid of it.

+1 for getting rid of it.

+1000 for instituting some kind of data gathering program within Fedora to get accurate data on what people download / instantiate / build upon from all the many binary editions of Fedora that get published.

+1 - Can we start to send out some communications around this so people won't be "surprised"?

Maybe file this as a self-contained change?

Discussed in the cloud meeting today. There are some arguments that there are some apps that are 32bit and don't support multilib (running on 64 bit). It was proposed that we have 32bit as a secondary arch.

I am fine with having it as a secondary arch but would prefer for it to be a short lived thing. Here is what I would like for it to mean:

• 1 or 2 cycles (f23/f24)
• only thing provided is 32 bit cloud image downloads (qcow2/raw)
• no uploads to public clouds.

Thoughts?

+1 to dusty's idea in #10.

Looks like we may have miscommunicated a bit. Looks like releng is no longer going to be building cloud images (see link).

https://pagure.io/releng/37943686086180fb95b4ff4b6e31192d63b5b391

We can add back the 32 bit images, but we will have to get oz fixed to enable them to build.

Yes, let's please add those back and get a comms plan going to sunset the image in the F24 timeframe. Any objections?

None from me.

jzb can you help us come up with a communication/marketing plan for this for F24 ?

@dustymabe indeed. Yes.

This seems to be pretty much wrapped up. Closing. Thanks all!