)]}'
{"/PATCHSET_LEVEL":[{"author":{"_account_id":4393,"name":"Dan Smith","email":"dms@danplanet.com","username":"danms"},"change_message_id":"ef7920d2a5c8f8973cc7ba535c87e2125f4c977c","unresolved":false,"context_lines":[],"source_content_type":"","patch_set":1,"id":"567a2d43_97695173","updated":"2024-09-24 18:34:19.000000000","message":"Just to capture what we discussed on IRC just now:\n\nI understand the desire to know what can be (expected to be) dd\u0027able to a disk on some sort of hardware. The problem is, that differs by context (machine, virtual or physical), etc. It would be trivial(ish) for a PC hardware manufacturer to teach their firmware to read and boot from a qcow2-formatted physical device, and it would be trivial (because it exists) for the kernel to be able to read such a device natively. I think the policy for \"what are we *willing* to dd to a disk\" might exist in ironic or nova and probably should be defined there.\n\nIf we need to have a policy in oslo, I think it should be a list of dd\u0027able formats and not a property on the inspectors themselves. However, we already know for a fact that nova and ironic\u0027s policies will be different, which I think means it should go into the projects themselves and not here in oslo.","commit_id":"1a9374c1170f0f47ab4ffe10f0e1983991da000f"},{"author":{"_account_id":10342,"name":"Jay Faulkner","display_name":"JayF","email":"jay@jvf.cc","username":"JayF","status":"youtube.com/@oss-gr / podcast.gr-oss.io"},"change_message_id":"23bfe2d7cb08a003dcc2cd5b07720d79333e9fec","unresolved":false,"context_lines":[],"source_content_type":"","patch_set":1,"id":"303d4c9a_405820bd","updated":"2024-09-24 17:19:29.000000000","message":"The motivation here is this:\n- GPT images used to be detected as \u0027raw\u0027\n- Draft Ironic code up for review was broken when these began being detected as \u0027gpt\u0027 instead.\n- Ironic clearly needs a signal from the inspector that an image is a \"raw-like\" image that can be written bit for bit.\n\nOther options include:\n- Ironic keeping this list internally, meaning we have to pin oslo.utils versions and/or worry about breakage anytime new inspectors are added\n- Keeping a list, similar to ALL_FORMATS, listing all formats that can be written out bit-for-bit\n\n\nWhy does Ironic care so much about if an image is raw?\n- If an image doesn\u0027t require conversion, in our most scalable configurations, we are able to write it to the disk in a streaming manner -- meaning at no point in time would we even have the whole image downloaded as a separate file or call qemu-img against it.","commit_id":"1a9374c1170f0f47ab4ffe10f0e1983991da000f"},{"author":{"_account_id":10342,"name":"Jay Faulkner","display_name":"JayF","email":"jay@jvf.cc","username":"JayF","status":"youtube.com/@oss-gr / podcast.gr-oss.io"},"change_message_id":"04b0ec532bab104ec43d88d0837d6348ff7977f9","unresolved":false,"context_lines":[],"source_content_type":"","patch_set":1,"id":"a2ea1d12_f7c13e85","in_reply_to":"567a2d43_97695173","updated":"2024-09-24 18:36:48.000000000","message":"Also noting that Dan suggested, and Ironic will likely be implementing it this way, that we can pass a list of inspectors when calling format_inspector, to prevent \"rugpull\" style breakages if/when new oslo.utils starts counting images Ironic would treat as \"raw\" as some other format.","commit_id":"1a9374c1170f0f47ab4ffe10f0e1983991da000f"}]}