Flush stor

[kostyanf14]

No description provided.

[akihikodaki]

https://oidref.com/1.3.6.1.4.1.318.1.1.12.3.3.1.1.4 says it has delayedReboot so you may check if the value is used. I don't require to do so as Microsoft's documentation does not mention it.

[benyamin-codez]

@kostyanf14

Thanks for the reference.

I noticed the MS doco suggested the following ASNs are required:

• ImmediatePowerOn = 1
• ImmediatePowerOff = 2
• ImmediateReboot = 3

[sic] - I'm presuming typographic errors here, and these should actually be:

• immediateOn = 1
• immediateOff = 2
• immediateReboot = 3

...from https://oidref.com/1.3.6.1.4.1.318.1.1.12.3.3.1.1.4 (the relevant OID):

rPDUOutletControlOutletCommand OBJECT-TYPE
   SYNTAX INTEGER {
      immediateOn             (1),
      immediateOff            (2),
      immediateReboot         (3),
      delayedOn               (4),
      delayedOff              (5),
      delayedReboot           (6),
      cancelPendingCommand    (7)
   }

As the first two, i.e. immediateOn and immediateOff, are read-write, i.e. set/get, perhaps we are performing the system-reset via QMP, not only when the test issues the command but perhaps also when the test issues a query to check the power state.

So when asked to check the power state, perhaps we are resetting instead... 8^O

This might explain the erratic results seen in virtio-win...
Perhaps we need to tee the socket data to a SNMP parser to check...
...but it might be quicker / easier to add the requisite features to the listener here... 8^d

[kostyanf14]

@benyamin-codez

No, you are not correct. We analyze requests, and HLK always sends ImmediateReboot and nothing else.

[kostyanf14]

@benyamin-codez Also, if you open HLKX results where Flush test failed, you can see the error. Error looks like this: System reset was requested but system not rebooted.

[benyamin-codez]

@kostyanf14

Thanks for the prompt reply.

No, you are not correct. We analyze requests, and HLK always sends ImmediateReboot and nothing else.

I saw the socket.recvfrom(0), with the zero-sized buffer, and presumed you weren't looking...

Also, if you open HLKX results where Flush test failed, you can see the error. Error looks like this:
System reset was requested but system not rebooted.

Oh, ok. I was checking Failed: Flush Test.zip > Test Run > WttEA.log instead
I've only seen 0x80070002 (ERROR_FILE_NOT_FOUND errors) in the dozen or so I've checked.

I'll grab the Studio and have a peek... 8^d
Can you tell I'm trying to avoid installing the whole HLK...? 8^D It seems like quite a beast...

[kostyanf14]

I saw the socket.recvfrom(0), with the zero-sized buffer, and presumed you weren't looking...

During development, we checked this data and found that it can be ignored.

For example:
virtio-win/kvm-guest-drivers-windows#1304 - viostor-Win2022x64

The error is not related to the storage driver but something with HLK.

[benyamin-codez]

Yeah, that's what I see in the few I peeked into.
The actual test passes and then an error occurs during test cleanup.
Apparently it occurs whilst "Initializing and formatting the disk"...
So it doesn't appear to be related to this SNMP kit or reboots at all - at least in these cases.

Is it possible the disk is OFFLINE and so it cannot be initialised or formatted?
... If so, this is something we might need to address in the driver...
Is it perhaps something to do with the type of disk backing, i.e. COW or RAW?
This appears to be running (and failing) on non-bootable disks (as expected - except the failing part)...
I presume there's no AV or anything else interfering with anything in the boot block...

Quite annoying...! 8^(

[benyamin-codez]

System reset was requested but system not rebooted.

Do you perhaps see that in QEMU logs..?

The 0x8007007E error (missing module), i.e.:
Caught std::exception: onecore\drivers\storage\storutil\disk.cpp(478)\Storage.Tests.FlushTest.dll!...
... would seem to indicate part of the HLK is missing or misconfigured.

Possibly also the test is wiping the disk where the module is expected...?!?!
As in format is called which wipes the disk where Storage.Tests.FlushTest.dll is...?
e.g. C:\HLK\JobsWorkingDir\Tasks\WTTJobRunDCAC4D33-DFF4-EF11-BD88-56000200DDDD\Storage.Tests.FlushTest.dll

The 0X80070002 errors are usually related to the trace being missing (that could also be a hint):
e.g. C:\HLK\JobsWorkingDir\Tasks\WTTJobRunDCAC4D33-DFF4-EF11-BD88-56000200DDDD\Te.wtl

Is the storage::FlushFUATest::TestCleanup task misconfigured with or hard-coded to C:\ or \\?\ or \Device.....?!?

[kostyanf14]

You can configure only the following options:

• IP IP Address of Remote PDU
• OID OID of Remote PDU outlet
• Outlet Port of Remote PDU outlet
• Community Community of Remote PDU (e.g. private)

All other configurations are done automatically by HLK (not AutoHCK)

[kostyanf14]

@benyamin-codez Let's continue in the discussion thread #608. This is more complicated to find this PR and this creates extra mail noise for other people.