Omapi makes DHCPD up to 100 % CPU Usage

[Shawn Routhier]

On 04/07/2011 07:16, Jiri Popelka wrote:
> On 02/23/2011 01:37 AM, Shawn Routhier wrote:
>> On 02/21/2011 08:26, Peter Rathlev wrote:
>>> On Mon, 2011-02-21 at 16:25 +0100, Jürgen Dietl wrote:
>>>> Operating System: SUSE Enterprise Server 11 (x86_64)
>>>> Version = 11
>>>> PatchLevel = 1
>>>>
>>>> ISC-DHCP Server 4.2.0-P2
>>>>
>>>> When I try to open an omapi port via the omshell or script the dhcpd
>>>> process raises to 100 % of CPU usage and when I try to connect over
>>>> omapi I get the message
>>>
>>> Redhat har the following bug entry on that:
>>>
>>> https://bugzilla.redhat.com/show_bug.cgi?id=666441
>>>
>>> The patch is "dhcp-4.2.0-P2-omapi.patch" which can be found in the
>>> source RPM. If you have trouble finding the patch (and trust a
>>> stranger) it's this:
>>>
>>> diff -up dhcp-4.2.0-P2/omapip/protocol.c.omapi
>>> dhcp-4.2.0-P2/omapip/protocol.c
>>> --- dhcp-4.2.0-P2/omapip/protocol.c.omapi 2009-11-20
>>> 02:49:02.000000000 +0100
>>> +++ dhcp-4.2.0-P2/omapip/protocol.c 2011-01-03 16:30:25.000000000 +0100
>>> @@ -386,7 +386,7 @@ isc_result_t omapi_protocol_signal_handl
>>> /* Should only receive these when opening the initial authenticator. */
>>> if (!strcmp (name, "status")) {
>>> status = va_arg (ap, isc_result_t);
>>> - if (status != ISC_R_SUCCESS) {
>>> + if ((status != ISC_R_SUCCESS)&& (status>= ISC_RESULTCLASS_DHCP)) {
>>> omapi_signal_in (h -> inner, "status", status,
>>> (omapi_object_t *)0);
>>> omapi_disconnect (p -> outer, 1);
>>>
>>> It's not in vanilla 4.2.0-P2. Beware that there might be a reason for
>>> this, though I wouldn't know any. We use the patch on our servers and
>>> haven't seen any problems.
>>>
>>
>> 4.2.1rc1 has now been released. It contains a different fix that should
>> address the underlying omapi issue.
>
> Hello Shawn,
>
> looking at the RELNOTES for Changes since 4.2.1b1 I'm not sure
> which of the fixes is the one you had in mind (there's no OMAPI mentioned).
> But I guess it's the one with DDNS handling ([ISC-Bugs #23103]).
> Can you confirm this, please ?
>
> Do you know if this problem could also occur with dhcp-4.1/3.1 ?
> Or it's dhcp-4.2.0 only (because of how dhcp-4.2.0 process transactions
> to DNS in an asynchronous fashion) ?
>
> Thank you very much for answers,
>
> Jiri Popelka
> Red Hat, inc.

I was referring to #21839 which was released in 4.2.1b1.

The issue for that ticket was introduced as part of the re-plumbing of
the connection code in 4.2.0 and wouldn't affect 4.1 or 3.1.  The
re-plumbing was required due to the DDNS changes but isn't specifically
for them.

As always if people are still seeing an issue in newer code they should
either open or re-open a ticket.

Shawn