See RFE 6219078

-- Renaud

Steffen Weiberle wrote:
> Renaud Manus wrote On 06/28/06 09:58,:
>>
>>
>> Steffen Weiberle wrote:
>>
>>> I've been looking all over and can't find this...
>>>
>>> In the past I have seen error messages when a service terminates too 
>>> often that says service restarted too many times and has been put 
>>> into maintenance mode. I even thought I had seen a parameter 
>>> somewhere that set it to 3 or so. Howver, now I can't find it.
>>
>>
>> This is hardcoded into usr/src/cmd/svc/startd/startd.h
>>
>> #define FAULT_THRESHOLD         3
>>
>>>
>>> The reason is that I have come across a situation/bug where vold 
>>> (volfs service) core dumps in 6/06. Vold is restarted every four to 
>>> six seconds, so I am getting about ten abnormal ends and resulting 
>>> core dumps a minute! Took me a while to figure out why my disk would 
>>> not stop :)
>>>
>>> Anyway, it seems SMF should catch the fact that this error is 
>>> happening much to frequently and put volfs into maintenance mode. But 
>>> it is not. So I have been searching for the variable, but can't find 
>>> it. I don't see it in the manifests for the services I have looked 
>>> at.  Where is it?
>>
>>
>> svc.startd(1M) man page says:
>>
>>       If three method failures happen in a row, or if the  service
>>       is restarting more than once a second, svc.startd places the
>>       service in the maintenance state.
> 
> Now I overlooked that!
> 
>>
>> It seems that in your case, the start method doesn't fail but the
>> service is restarting every four or six seconds hence the rules don't
>> apply and the instance never goes into maintenance.
> 
> As I was trawling through messages on this topic I did see some 
> references to why/why not put a service into maintenance. One comment 
> was that it the same error happens over and over, what is the purpose of 
> retrying. This seems like a perfect case where retries should be limited.
> 
> Off-hand, is this type of issue being addressed via a CR for bug or RFE? 
> If not, I can file one.
> 
> Thanks
> Steffen
> 
>>
>> -- Renaud
> _______________________________________________
> smf-discuss mailing list
> smf-discuss at opensolaris.org

Reply via email to