Re: [Xen-devel] [OSSTEST PATCH RFC v2 11/14] Introduce ts-xtf-run

[Ian Jackson <ian.jackson@xxxxxxxxxxxxx>]

Andrew Cooper writes ("Re: [Xen-devel] [OSSTEST PATCH RFC v2 11/14] Introduce 
ts-xtf-run"):
> On 08/08/16 10:22, Wei Liu wrote:
...
> > The script may exit early if it detects the test host is down or
> > xtf-runner returns non-recognisable exit code.
> 
> What is OSSTests general approach to recovering crashed hosts?

osstest will attempt to recover the host by power cycling it, for the
purposes of capturing logs.

It won't resume the test job where it left off.  I don't think that's
desirable.

> XenRT recovers the host and continues testing, so subsequent steps are
> not blocked.  (especially in a case like this, where there is a
> distinctly more likely chance that the host might crash.)

The effect of Wei's patch is that, if one of the XTF tests causes the
host to crash:
 - this will be treated as a failure of that specific XTF test
- the osstest log capture will initially fail and the host will be
    power cycled in an attempt to be able to capture logs stored on
    the host hard disk
 - the remaining XTF tests in that job will not be run
 - the other jobs in the same flight will run (and perhaps experience
    the same bug, or perhaps not)
 - the next flight to allocate the crashed host will wipe it by
    power cycling it and netbooting debian-installer

Ian.

_______________________________________________
Xen-devel mailing list
Xen-devel@xxxxxxxxxxxxx
https://lists.xen.org/xen-devel