Hans,
I tested the repair command, it worked. Can you please commit the patch.
Thanks
-Nagu
-----Original Message-----
From: Hans Feldt [mailto:[email protected]]
Sent: 02 May 2013 16:30
To: Nagendra Kumar
Cc: [email protected]
Subject: Re: [devel] [PATCH 1 of 1] Amf: Mark SU oper state disable during its
state change from instantiating to term fail (#3119)
On 05/02/2013 06:57 AM, Nagendra Kumar wrote:
> Hi,
> As I said before, there is no handling of term fail in Amf code as per
> Specs. This change could bring the first one, could lead some inconsistency.
> Ok with that ?
Yes, I am not OK with failing over something that is in unknown state.
That surely is totally against the spec and the intention of AMF?
Can you try mmy suggested approach and make sure you can repair the SU?
Thanks,
Hans
>
> Thanks
> -Nagu
>
> -----Original Message-----
> From: Hans Feldt [mailto:[email protected]]
> Sent: 30 April 2013 17:24
> To: Nagendra Kumar
> Cc: [email protected]
> Subject: RE: [devel] [PATCH 1 of 1] Amf: Mark SU oper state disable
> during its state change from instantiating to term fail (#3119)
>
>> -----Original Message-----
>> From: Nagendra Kumar [mailto:[email protected]]
>> Sent: den 30 april 2013 13:51
>> To: Hans Feldt
>> Cc: [email protected]
>> Subject: RE: [devel] [PATCH 1 of 1] Amf: Mark SU oper state disable
>> during its state change from instantiating to term fail (#3119)
>>
>> But if we have disablerestart implemented, then it may have problem.
>
> Should be no problem as long as a component has well behaving CLC CLI
> scripts...
>
>>
>> As per Specs:
>> In this case, the presence state of the component is not set to
>> restarting but transitions through the other presence state values
>> (typically in the absence of failures: terminating, uninstantiated,
>> instantiating, and then instantiated) as the component is terminated and
>> instantiated again.
>>
>> In this also, the component can go into term failed from
>> instantiating. But this may require su failover.
>
> If a component is in TERM-FAILED AMF can never do SU failover. That would be
> a bug.
> /Hans
>
>>
>> Thanks
>> -Nagu
>>
>> -----Original Message-----
>> From: Hans Feldt [mailto:[email protected]]
>> Sent: 29 April 2013 18:32
>> To: Nagendra Kumar
>> Cc: [email protected]
>> Subject: Re: [devel] [PATCH 1 of 1] Amf: Mark SU oper state disable
>> during its state change from instantiating to term fail (#3119)
>>
>> Here's what I suggest instead:
>>
>>> diff --git a/osaf/services/saf/avsv/avnd/avnd_susm.c
>>> b/osaf/services/saf/avsv/avnd/avnd_susm.c
>>> --- a/osaf/services/saf/avsv/avnd/avnd_susm.c
>>> +++ b/osaf/services/saf/avsv/avnd/avnd_susm.c
>>> @@ -1435,6 +1435,18 @@ uint32_t avnd_su_pres_st_chng_prc(AVND_C
>>> goto done;
>>> }
>>>
>>> + /* instantiating -> term-failed */
>>> + if ((SA_AMF_PRESENCE_INSTANTIATING == prv_st) &&
>>> + (SA_AMF_PRESENCE_TERMINATION_FAILED
>>> + == final_st))
>> {
>>> + TRACE("SU Instantiating -> Termination Failed");
>>> + m_AVND_SU_OPER_STATE_SET(su,
>> SA_AMF_OPERATIONAL_DISABLED);
>>> + m_AVND_SEND_CKPT_UPDT_ASYNC_UPDT(cb, su,
>> AVND_CKPT_SU_OPER_STATE);
>>> + /* Don't send su-oper state msg, just update su
>>> oper state
>>> + * AMF has lost control over this component
>>> + and the operator
>> needs
>>> + * to repair this node. Failover is not possible in
>>> this state. */
>>> + avnd_di_uns32_upd_send(AVSV_SA_AMF_SU,
>> saAmfSUOperState_ID, &su->name, su->oper);
>>> + }
>>> +
>>> /* instantiated/restarting -> inst-failed */
>>> if (((SA_AMF_PRESENCE_INSTANTIATED == prv_st) ||
>> (SA_AMF_PRESENCE_RESTARTING == prv_st)) &&
>>> (SA_AMF_PRESENCE_INSTANTIATION_FAILED ==
>>> final_st)) {
>>
>>
>> what do you think?
>>
>> /Hans
>>
>> On 04/24/2013 11:37 AM, Nagendra Kumar wrote:
>>>>> Yes in that case there can be no SU failover. It is not AMFs fault
>>>>> but the
>> component that has a bad CLC CLI script.
>>> Not implemented. :)
>>>
>>> Is this Ack with recvr as 0 ?
>>>
>>> -Nagu
>>>
>>> -----Original Message-----
>>> From: Hans Feldt [mailto:[email protected]]
>>> Sent: 24 April 2013 14:52
>>> To: Nagendra Kumar; [email protected]
>>> Cc: [email protected]
>>> Subject: RE: [devel] [PATCH 1 of 1] Amf: Mark SU oper state disable
>>> during its state change from instantiating to term fail (#3119)
>>>
>>>> -----Original Message-----
>>>> From: Nagendra Kumar [mailto:[email protected]]
>>>> Sent: den 24 april 2013 11:18
>>>> To: Hans Feldt; [email protected]
>>>> Cc: [email protected]
>>>> Subject: RE: [devel] [PATCH 1 of 1] Amf: Mark SU oper state disable
>>>> during its state change from instantiating to term fail (#3119)
>>>>
>>>> It is a case of instantiating to term failed. So, I don't find any
>>>> problem in sending 0 or su failover.
>>>>
>>>> Anyway, what should be the scenario when Act Su (2N case) goes into
>>>> term failed state and if we don't do su failover, only Standby SU
>>>> remains. Is this the case you are talking? We are not following it
>>>> in any red model. We are
>>>
>>> Yes in that case there can be no SU failover. It is not AMFs fault
>>> but the
>> component that has a bad CLC CLI script.
>>> /Hans
>>>
>>>> removing(graceful) susi of faulted SU and we are doing su failover.
>>>>
>>>> Thanks
>>>> -Nagu
>>>>
>>>> -----Original Message-----
>>>> From: Hans Feldt [mailto:[email protected]]
>>>> Sent: 24 April 2013 14:34
>>>> To: Nagendra Kumar; [email protected]
>>>> Cc: [email protected]
>>>> Subject: RE: [devel] [PATCH 1 of 1] Amf: Mark SU oper state disable
>>>> during its state change from instantiating to term fail (#3119)
>>>>
>>>> But in term-failed state you cannot move anything to other SUs. The
>>>> node in questions needs to be repaired first. So maybe you should
>>>> just set the oper state to disable but not use the
>>>> avnd_di_oper_send() which signals to amfd to actually do something?
>>>> Thanks,
>>>> Hans
>>>>
>>>>> -----Original Message-----
>>>>> From: Nagendra Kumar [mailto:[email protected]]
>>>>> Sent: den 24 april 2013 11:05
>>>>> To: Hans Feldt; [email protected]
>>>>> Cc: [email protected]
>>>>> Subject: RE: [devel] [PATCH 1 of 1] Amf: Mark SU oper state
>>>>> disable during its state change from instantiating to term fail
>>>>> (#3119)
>>>>>
>>>>> When SU goes to disable state(thereby Out of Service), we are
>>>>> moving assignments to other healthy SUs if there is no any
>>>>> redundancy
>> constraints.
>>>>> At Amfd, we are not checking recovery until reported as node f/s over.
>>>>>
>>>>>
>>>>> Thanks
>>>>> -Nagu
>>>>>
>>>>> -----Original Message-----
>>>>> From: Hans Feldt [mailto:[email protected]]
>>>>> Sent: 24 April 2013 13:27
>>>>> To: Nagendra Kumar; [email protected]
>>>>> Cc: [email protected]
>>>>> Subject: RE: [devel] [PATCH 1 of 1] Amf: Mark SU oper state
>>>>> disable during its state change from instantiating to term fail
>>>>> (#3119)
>>>>>
>>>>> I think in the term-failed state the rcvr parameter to
>>>>> avnd_di_oper_send() should be 0 and not SU-FAILOVER as it will be
>>>>> with this change. Because in term-failed SU-FAILOVER cannot be
>> allowed.
>>>>>
>>>>> Seems like you need to create a new if statement or so.
>>>>>
>>>>> Thanks,
>>>>> Hans
>>>>>
>>>>>> -----Original Message-----
>>>>>> From: [email protected] [mailto:[email protected]]
>>>>>> Sent: den 24 april 2013 09:19
>>>>>> To: [email protected]
>>>>>> Cc: [email protected]
>>>>>> Subject: [devel] [PATCH 1 of 1] Amf: Mark SU oper state disable
>>>>>> during its state change from instantiating to term fail (#3119)
>>>>>>
>>>>>> osaf/services/saf/avsv/avnd/avnd_susm.c | 5 +++--
>>>>>> 1 files changed, 3 insertions(+), 2 deletions(-)
>>>>>>
>>>>>>
>>>>>> diff --git a/osaf/services/saf/avsv/avnd/avnd_susm.c
>>>>>> b/osaf/services/saf/avsv/avnd/avnd_susm.c
>>>>>> --- a/osaf/services/saf/avsv/avnd/avnd_susm.c
>>>>>> +++ b/osaf/services/saf/avsv/avnd/avnd_susm.c
>>>>>> @@ -1425,8 +1425,9 @@ uint32_t
>> avnd_su_pres_st_chng_prc(AVND_C
>>>>>> }
>>>>>>
>>>>>> /* instantiating -> inst-failed */
>>>>>> - if ((SA_AMF_PRESENCE_INSTANTIATING == prv_st) &&
>>>>>> (SA_AMF_PRESENCE_INSTANTIATION_FAILED == final_st)) {
>>>>>> - TRACE("SU Instantiating -> Instantiation
>>>>>> Failed");
>>>>>> + if ((SA_AMF_PRESENCE_INSTANTIATING == prv_st) &&
>>>>>> ((SA_AMF_PRESENCE_INSTANTIATION_FAILED == final_st) ||
>>>>>> +
>>>>>> (SA_AMF_PRESENCE_TERMINATION_FAILED== final_st))) {
>>>>>> + TRACE("SU Instantiating ->
>>>>>> Instantiation/Termimation
>>>>>> Failed");
>>>>>> /* send the su-oper state msg (to indicate that
>>>>> instantiation
>>>>>> failed) */
>>>>>> m_AVND_SU_OPER_STATE_SET(su,
>>>>>> SA_AMF_OPERATIONAL_DISABLED);
>>>>>> m_AVND_SEND_CKPT_UPDT_ASYNC_UPDT(cb, su,
>>>>> AVND_CKPT_SU_OPER_STATE);
>>>>>>
>>>>>> -----------------------------------------------------------------
>>>>>> -
>>>>>> --
>>>>>> --
>>>>>> -------- Try New Relic Now & We'll Send You this Cool Shirt New
>>>>>> Relic is the only SaaS- based application performance monitoring
>>>>>> service that delivers powerful full stack analytics. Optimize and
>>>>>> monitor your browser, app, & servers with just a few lines of code.
>>>>>> Try New Relic and get this awesome Nerd Life shirt!
>>>>>> http://p.sf.net/sfu/newrelic_d2d_apr
>>>>>> _______________________________________________
>>>>>> Opensaf-devel mailing list
>>>>>> [email protected]
>>>>>> https://lists.sourceforge.net/lists/listinfo/opensaf-devel
>>>
>>>
>
>
------------------------------------------------------------------------------
This SF.net email is sponsored by Windows:
Build for Windows Store.
http://p.sf.net/sfu/windows-dev2dev
_______________________________________________
Opensaf-devel mailing list
[email protected]
https://lists.sourceforge.net/lists/listinfo/opensaf-devel
