hello all, does illumos supports MegaRAID 9560-8i ?

hello all, may I know which one could help to take a look for 9560-8i driver issue? pastebin.com/uUr3qp9g ; I have post this question in smartos and illumos channel, thanks in advance

Got ipv6 working in both zones, lx and kvm(thanks bahamat!) like a charm which is great. However, global and non-global zones now all share the same IPv6-address. Is there a way to assign each vm/zone its own static IPv6-address in the delegated prefix instead of sharing one for all? Tried changing "addrconf" to a "2a02-address" in the ips-array but it keeps complaining its not right.

Did the same in "allowed_ips", same result.

Also tried "ipadm create-addr -T static -a <address> <object name>" but got "permission denied" or "from-gz"-error

I just tried netbooting the latest smartos image, I have the same results fetches the files then is stuck and then again back to the efi shell

tozhu what's the output of prtconf -vp ?

I think the url is not valid anymore as it points to joyent on the pxe menu pasteboard.co/sood7AYmlBhZ.png

This is the last that I see on netboot after this is back to the efi shell pasteboard.co/rN3gYbYWrYWh.png

how big is the boot archive?

tsoome_ it seems it downloaded ok according to the messages, the size of the boot archive is around 60mb

if you can get unix, boot_archive and boot_archive.hash loaded into memory, but the actual boot fails, there can be only few reasons, one is that switching off EFI boot services fails (I assume ipxe will exit on that), or relocating loaded files fail (not enough memory below 4GB or not large enough chunk to store either unix or boot_archive).

60MB is probably compressed, it will be uncompressed during the read, but still not too huge.

with loader you could just load them and use lsmod -v to see the sizes

also you could check memory map. I do not know ipxe to suggest diag/debug commands.

but the fact that you end up on EFI shell prompt hints that there was an error and ipxe did exit

tsoome_ thanks I'll check that

Hi, is there is a Triton CNAPI call I make to list all the different traits on all CNS?

Disregard. This worked: `sdc-cnapi /servers | json -Ha hostname traits`

tsoome_ There is no much debug options in pxe, I just tried chain --autofree but same result could be something related to oracle cloud but the linux pxe boot work

well, sure it will. so should ours, but first we would need to understand, why it is failing:)

unfortunately I failed to create my "free" account there, will try a bit later...

tsoome_ what I did to get to the efi shell was just create an ubuntu free instance, then rename the EFI folder to something else , but first you will need to download a patched bin netboot.smartos.org/ipxe.efi

That one could boot smartos os as netboot upstream broke multiboot2 implementation bahamat has more details

we have one issue with our mb2, see code.illumos.org/c/illumos-gate/+/2292

that is, we assume we can use memmap, but UEFI boot loader may not provide it.

The thing is that seems everybody is using netboot for smartos on triton I think, so this issue maybe is just isolated to oracle cloud

very likely

it would still be nice to have it fixed

neirac: The boot archive is 317MB. If you're only getting 60, then something is disrupting your session.

Bahamat oh, I just tried a browser to download it to check the size maybe I cut it early

Even compressed, it's >180MB.

And the joyent URLs in the netboot server are still valid. They're cnamed to the real domain.

bahamat that should be a problem I see that the hash if checked pasteboard.co/rN3gYbYWrYWh.png seems an isolated issue if it's working on your side, but I don't know how to debug this on my side, maybe build the patched pxe with debug message?

Maybe.

where is the source for ipxe.efi?, is on the tritondatacenter repo?

Yes

this one right ? github.com/TritonDataCenter/ipxe

I'm not sure I would call this a bug, but if you dd zeros to a drive that is not in use but for which the kernel is aware of the disk label, you get endless messages about the primary disk label corrupted so using secondary, and the box stops responding on the console and network all together

(I ended up hard resetting it)

I'm googling how to make illumos ignore the disk labels, in case I need to zero a drive in illumos

More specifically - learning the subsystem

bahamat which image uuid are you using to build ipxe? running make expects a certain platform and image to run the build

I'm using e6539349-b85c-4401-92ef-ea4dad86c2bc but should be fine is just for testing where it fails