Dan Kenigsberg has posted comments on this change. Change subject: network: wait for udev after configuring SRIOV ......................................................................
Patch Set 5: (1 comment) https://gerrit.ovirt.org/#/c/40400/5/vdsm/network/api.py File vdsm/network/api.py: Line 542: # This is a blocking call that should wait for all udev events to be Line 543: # handled. Since those include renaming devices, we should wait until Line 544: # udev finishes in order to report a stable hardware state in Line 545: # subsequent engine calls. Line 546: udevadm.settle(timeout=10) > I sent a patch for udev: If "settle" is not to be used in new code, it must be deprecated, explained and documented. Do you know where Lenart's assertion comes from? From this inherent race between injecting the events into udev and running settle? We have a similar race in ifup, which may return before addresses are available, and we plant to solve by monitoring netlink events. We can similarly use `udevadm monitor` to solve this race here. Line 547: Line 548: Line 549: def _persist_numvfs(device_name, numvfs): Line 550: dir_path = os.path.join(constants.P_VDSM_LIB, 'virtual_functions') -- To view, visit https://gerrit.ovirt.org/40400 To unsubscribe, visit https://gerrit.ovirt.org/settings Gerrit-MessageType: comment Gerrit-Change-Id: I04d63ca35aab585051c3d0a5f2652dbf2f91b080 Gerrit-PatchSet: 5 Gerrit-Project: vdsm Gerrit-Branch: master Gerrit-Owner: Ido Barkan <ibar...@redhat.com> Gerrit-Reviewer: Dan Kenigsberg <dan...@redhat.com> Gerrit-Reviewer: Ido Barkan <ibar...@redhat.com> Gerrit-Reviewer: Jenkins CI Gerrit-Reviewer: Martin Polednik <mpoled...@redhat.com> Gerrit-Reviewer: Nir Soffer <nsof...@redhat.com> Gerrit-Reviewer: automat...@ovirt.org Gerrit-HasComments: Yes _______________________________________________ vdsm-patches mailing list vdsm-patches@lists.fedorahosted.org https://lists.fedorahosted.org/mailman/listinfo/vdsm-patches