On Sun, Apr 02, 2023 at 12:36:41AM +0200, Miroslav Suchý wrote: > Dne 01. 04. 23 v 20:27 Kevin Fenzi napsal(a): > > Should we stop uploading 'standard' and 'gp2'? I mean, is there any > > reason anyone would want to use one of those? I find it just confusing > > that there's multiple types for each image. If we can, I'd say we should > > just do gp3? > > standard is magnetic volume of previous generation > > https://docs.aws.amazon.com/AWSEC2/latest/UserGuide/ebs-volume-types.html#vol-type-prev > > Yes, this is obsolete and should be replaced by either st1 (better throughput) or sc1. > > They are super cheap. E.g., sc1 cost $0.015 (compare to gp3 with $0.080). > And IMHO st1 or sc1 has sense for rootfs volumes on machines where > everything is loaded in memory after start. Or for data that are rarely > accessed. ok. Perhaps this should be a question for the cloud sig? > The gp2->gp3 is worth doing always. You just need to mind the settings. If > the volume is over 1TB you should setup throughput and IOPS over the > baseline to get same performance. But even for 6TB volumes you get better > price for gp3. See the table at "Cost comparison between gp2 and gp3 in the > us-east-1 (N. Virginia) Region" at the bottom of > > https://aws.amazon.com/blogs/storage/migrate-your-amazon-ebs-volumes-from-gp2-to-gp3-and-save-up-to-20-on-costs/ > > And even for 16TB volumes with maximalized IOPS and Throughput you get 15% savings. > > You can play with the calculator https://aws.amazon.com/ebs/resources/ (xls sheet) Cool. > > How can you convert an existing machine? Is that something on the aws > > web console? or something via the cli? We should definitely convert all > > our instances. :) > > "How to migrate from gp2 to gp3" chapter from > > https://aws.amazon.com/blogs/storage/migrate-your-amazon-ebs-volumes-from-gp2-to-gp3-and-save-up-to-20-on-costs/ > > You can do this both from WebUI and from command line. Great! > > Finally, on timing... we should wait until FESCo approves the change > > (but I don't see why they wouldn't). > > We do not need to. The change is about default storage of Fedora Cloud images. It is not related to what we use in infra :) > > Of course we can wait and align if we want to. fedimg is the thing that uploads the official cloud images, so if it does gp3, then the fedora images are gp3. :) > > We do go into f38 final freeze on > > tuesday next week, but we can try and change it monday? Or just get a > > freeze break to push it out after that. > > Freeze is next week? Who stole the time? I wish I knew. :) > OK, I think there is no need to rush. > > I will migrate Copr machines at the begging of the week. Likely on Monday > after I discuss it with the team. And Copr is not in the set of machine that > we freeze anyway. > > And after un-freeze I can migrate the rest. Or we can wait till Fesco approves the change and allign with it. Sounds great! Thanks for finding all this out. kevin
Attachment:
signature.asc
Description: PGP signature
_______________________________________________ infrastructure mailing list -- infrastructure@xxxxxxxxxxxxxxxxxxxxxxx To unsubscribe send an email to infrastructure-leave@xxxxxxxxxxxxxxxxxxxxxxx Fedora Code of Conduct: https://docs.fedoraproject.org/en-US/project/code-of-conduct/ List Guidelines: https://fedoraproject.org/wiki/Mailing_list_guidelines List Archives: https://lists.fedoraproject.org/archives/list/infrastructure@xxxxxxxxxxxxxxxxxxxxxxx Do not reply to spam, report it: https://pagure.io/fedora-infrastructure/new_issue
