Re: [ovirt-devel] vdsm-client getHardwareInfo takes X4 than vdsClient getVdsHardwareInfo

Avihai, can you file a bug for this, and link to https://bugzilla.redhat. com/show_bug.cgi?id=1381899#c17 ? On Thu, Jun 29, 2017 at 5:36 PM Piotr Kliczewski <pkliczew(a)redhat.com&gt; wrote: > We could optimize it for one time use. Is there a BZ to track it? > > 29 cze 2017 15:57 "Nir Soffer" <nsoffer(a)redhat.com&gt; napisał(a): > >> On Thu, Jun 29, 2017 at 3:54 PM Avihai Efrat <aefrat(a)redhat.com&gt; wrote: >> >>> Hi Guys , >>>>> >>>>> In ovirt 4.2 vdsClient is deprecated so using vdsm-client in tests I >>>>> get timeout failures. >>>>> >>>> >>>>> When I tested both utilities on a 4.1 host I noticed vdsm-client >>>>> takes X4 than vdsClient . >>>>> >>>>> Is this known ? >>>>> >>>>> can we make it faster ? >>>>> >>>>> *Taken from client CLI :* >>>>> [root@storage-ge3-vdsm1 ~]# time vdsClient -s 0 getVdsHardwareInfo >>>>> systemFamily = 'Red Hat Enterprise Linux' >>>>> systemManufacturer = 'Red Hat' >>>>> systemProductName = 'RHEV Hypervisor' >>>>> systemSerialNumber = '4C4C4544-0053-5410-8047-B9C04F465931' >>>>> systemUUID = '07FD09C7-8461-4981-B859-A40C548E10FF' >>>>> systemVersion = '7.2-9.el7_2.1' >>>>> >>>>> *real 0m0.382s* >>>>> user 0m0.272s >>>>> sys 0m0.056s >>>>> >>>>> [root@storage-ge3-vdsm1 ~]# time vdsm-client Host getHardwareInfo >>>>> { >>>>> "systemProductName": "RHEV Hypervisor", >>>>> "systemSerialNumber": "4C4C4544-0053-5410-8047-B9C04F465931", >>>>> "systemFamily": "Red Hat Enterprise Linux", >>>>> "systemVersion": "7.2-9.el7_2.1", >>>>> "systemUUID": "07FD09C7-8461-4981-B859-A40C548E10FF", >>>>> "systemManufacturer": "Red Hat" >>>>> } >>>>> >>>>> *real 0m1.208s* >>>>> user 0m0.966s >>>>> sys 0m0.111s >>>>> >>>>> >> The difference is about 0.7 seconds. This can be explained by the >> time needed to load the yaml schema - we load it for every request, >> for validating the the request and generating online help. >> >> It takes about 0.1 seconds on a i7-4770 CPU @ 3.40GHz, but maybe you are >> testing on a much slower machine, or your machine is overloaded for some >> other reason? >> >> This was discussed in >> https://bugzilla.redhat.com/show_bug.cgi?id=1381899#c17 >> >> We can make this 100 times faster by using pickle format instead of >> parsing >> yaml. >> >> We can also make it infinitely faster by loading the schema only when >> generating online help. There is no real need to validate the request >> on the client side when the server side is already doing this. >> >> Nir >> >>