On 08/09/2017 03:02 PM, Daniel P. Berrange wrote: > On Wed, Aug 09, 2017 at 02:55:36PM +0200, Michal Privoznik wrote: >> On 08/09/2017 02:14 PM, Daniel P. Berrange wrote: >>> On Wed, Aug 09, 2017 at 02:00:06PM +0200, Michal Privoznik wrote: >>>> https://bugzilla.redhat.com/show_bug.cgi?id=1467245 >>>> >>>> Currently, there's a bug when undefining a domain with NVRAM >>>> store. Basically, the unlink() of the NVRAM store file happens >>>> during the undefine procedure iff domain is inactive. So, if >>>> domain is running and undefine is called the file is left behind. >>>> It won't be removed in the domain cleanup process either >>>> (qemuProcessStop). To avoid this forbid undefining domain with >>>> NVRAM file. >>> >>> Why do we need to forbid it ? Even if QEMU still has an open >>> file handle, it can continue to write to it after we unlink >>> it. >>> >>> >> >> That's what my v1 does. Anyway, there's third option: just recently >> Jirka added possibility to do some actions when domain is destroyed. He >> needed it for some migration work, but the design is broad enough to fit >> this problem too. What we can do is: >> >> if (flags & VIR_DOMAIN_UNDEFINE_NVRAM): >> if domain is running: >> register the callback /* that merely just unlinks the file */ >> else: >> unlink >> else: >> if (!(flags & VIR_DOMAIN_UNDEFINE_KEEP_NVRAM)): >> error >> >> >> What do you guys think of this one? > > The callback will be lost if libvirtd restarts. Ah, good point. In that case all I need is an ACK to v1 then ;-) Michal -- libvir-list mailing list libvir-list@xxxxxxxxxx https://www.redhat.com/mailman/listinfo/libvir-list
