7:23 a.m.
On 10/10/18 12:53 PM, Marcin Sobczyk wrote:
On 10/10/18 9:57 AM, Edward Haas wrote:
>
>
> On Mon, Oct 8, 2018 at 3:41 PM Marcin Sobczyk <msobczyk(a)redhat.com
> <mailto:msobczyk@redhat.com>> wrote:
>
> Hi,
>
> when I was working on BZ#1624306 it turned out, that even if I'd
> fix the help-printing code in vdsm-client, there's still a lot of
> inconsistencies in API's schema files.
>
> After some investigation, I found that there's a code for
> reporting schema inconsistencies, but it's turned off because it
> bloats the logs. There's also a "strict" mode that raises an
> exception each time an inconsistency is found, but it's also off
> because there's so many of them.
>
> I think that no one wants to maintain both: the actual code and
> schema files. My idea would be to make the schema derived from
> the code at build time as a long-term effort. Before we can do
> that though, we need to address the inconsistencies first.
>
> The schema is supposed to be the specification and the code its
> implementation. Generating the schema from the code does not sound
> right to me.
>
>
https://gerrit.ovirt.org/#/q/status:open+project:vdsm+branch:master+topic...
>
> This is a series of patches that helps fixing schema issues. What
> it does, is it adds a bunch of information from stack to vastly
> ease the process of fixing them. The logging level is also
> changed from 'WARNING' to 'DEBUG'. Here's an example of a
logged
> entry:
>
> Host.setKsmTune
> With message: Required property pages_to_scan is not provided
> when calling Host.setKsmTune
> With backtrace: [
> [
> "_verify_object_type",
> {
> "call_arg_keys":[
> "merge_across_nodes",
> "run"
> ],
> "schema_type_name":"KsmTuneParams"
> }
> ],
> [
> "_verify_complex_type",
> {
> "schema_type_type":"object"
> }
> ]
> ]
>
> To make it work, a patch for OST's 'basic-master-suite' is on the
> way that switches 'schema.inconsistency' logger's logging level
> do 'DEBUG'. That way, we can find all reported errors in
> 'vdsm-schema-error.log' files.
>
> Will we be able to detect the caller?
I will look at it.
I can include the context string in the log. Example entries:
With context string: from=::ffff:10.37.128.217,49550, flow_id=22456882
With context string: from=::1,51550
AFAIK you can trace back the caller by the 'flow_id'. Would that kind of
information suffice?
> An initial report that groups reported errors by
> namespaces/methods has also been made. Please note, that an
> implementation that completely avoids error duplicates is really
> hard and IMHO not worth the effort. You can find the results here:
>
> * Host: https://pastebin.com/YM2XnvTV
>
> We will need to cover the network stuff in here, thanks.
>
> * Image: https://pastebin.com/GRc1yErL
> * LVMVolumeGroup: https://pastebin.com/R1276gTu
> * SDM: https://pastebin.com/P3tfYEDD
> * StorageDomain: https://pastebin.com/Q0DxKgF5
> * StoragePool: https://pastebin.com/VXFWpfRC
> * VM: https://pastebin.com/tDC60c29
> * Volume: https://pastebin.com/TYkr9Zsd
> * |jobs|: https://pastebin.com/jeXpYyz9
> * |virt|: https://pastebin.com/nRREuEub
>
> Regards, Marcin
>
> _______________________________________________
> Devel mailing list -- devel(a)ovirt.org <mailto:devel@ovirt.org>
> To unsubscribe send an email to devel-leave(a)ovirt.org
> <mailto:devel-leave@ovirt.org>
> Privacy Statement: https://www.ovirt.org/site/privacy-policy/
> oVirt Code of Conduct:
> https://www.ovirt.org/community/about/community-guidelines/
> List Archives:
>
https://lists.ovirt.org/archives/list/devel@ovirt.org/message/NW67DZSIMQT...
>