[ovs-discuss] ovn-controller is taking 100% CPU all the time in one deployment
zhouhan at gmail.com
Thu Aug 29 19:34:12 UTC 2019
On Thu, Aug 29, 2019 at 12:16 PM Numan Siddique <nusiddiq at redhat.com> wrote:
> On Fri, Aug 30, 2019 at 12:37 AM Han Zhou <zhouhan at gmail.com> wrote:
>> On Thu, Aug 29, 2019 at 11:40 AM Numan Siddique <nusiddiq at redhat.com>
>> > Hello Everyone,
>> > In one of the OVN deployments, we are seeing 100% CPU usage by
>> ovn-controllers all the time.
>> > After investigations we found the below
>> > - ovn-controller is taking more than 20 seconds to complete full loop
>> (mainly in lflow_run() function)
>> > - The physical switch is sending GARPs periodically every 10 seconds.
>> > - There is ovn-bridge-mappings configured and these GARP packets
>> reaches br-int via the patch port.
>> > - We have a flow in router pipeline which applies the action - put_arp
>> > if it is arp packet.
>> > - ovn-controller pinctrl thread receives these garps, stores the
>> learnt mac-ips in the 'put_mac_bindings' hmap and notifies the
>> ovn-controller main thread by incrementing the seq no.
>> > - In the ovn-controller main thread, after lflow_run() finishes,
>> pinctrl_wait() is called. This function calls - poll_immediate_wake() as
>> 'put_mac_bindings' hmap is not empty.
>> > - This causes the ovn-controller poll_block() to not sleep at all and
>> this repeats all the time resulting in 100% cpu usage.
>> > The deployment has OVS/OVN 2.9. We have back ported the pinctrl_thread
>> > Some time back I had reported an issue about lflow_run() taking lot of
>> time -
>> > I think we need to improve the logical processing sooner or later.
>> > But to fix this issue urgently, we are thinking of the below approach.
>> > - pinctrl_thread will locally cache the mac_binding entries (just like
>> it caches the dns entries). (Please note pinctrl_thread can not access the
>> SB DB IDL).
>> > - Upon receiving any arp packet (via the put_arp action),
>> pinctrl_thread will check the local mac_binding cache and will only wake up
>> the main ovn-controller thread only if the mac_binding update is required.
>> > This approach will solve the issue since the MAC sent by the physical
>> switches will not change. So there is no need to wake up ovn-controller
>> main thread.
>> > In the present master/2.12 these GARPs will not cause this 100% cpu
>> loop issue because incremental processing will not recompute flows.
>> > Even though the above approach is not really required for master/2.12,
>> I think it is still Ok to have this as there is no harm.
>> > I would like to know your comments and any concerns if any.
>> > Thanks
>> > Numan
>> Hi Numan,
>> I think this approach should work. Just to make sure, to update the cache
>> efficiently (to avoid another kind of recompute), it should use ovsdb
>> change-tracking to update it incrementally.
>> Regarding master/2.12, it is not harmful except that it will add some
>> more code and increase memory footprint. For our current use cases, there
>> can be easily 10,000s mac_bindings, but it may still be ok because each
>> entry is very small. However, is there any benefit for doing this in
> I don't see much benefit. But I can't submit a patch to branch 2.9 without
> the fix getting merged in master first right ?
> May be once it is merged in branch 2.9, we can consider to delete it ?
> I think it is just about how would you maintain a downstream branch. Since
it is downstream, I don't think you need a change to be in upstream before
fixing a problem. In this case it may be *no harm*, but what if the
upstream is completely changed and incompatible for such a fix any more? It
shouldn't prevent you from fixing your downstream. (Of course it is better
to not have downstream at all, but sometimes it is useful to have it for a
temporary period, and since you (and us, too) are already there ... :)
-------------- next part --------------
An HTML attachment was scrubbed...
More information about the discuss