SASL - duplicate alarms bug

[Serge Aleynikov]

Nevertheless, there is a leftover dependency - the OS_MON's mib 
references EVA's mib, and therefore I should say that if you want to 
take advantage of OS_MON's alarms & SNMP you need EVA until that 
dependency is removed.

I ended up implementing functionality similar to the one in EVA where 
alarms can be mapped to SNMP traps, without EVA's overhead.

I wonder how others use alarms without use of SNMP...

Serge

Francesco Cesarini (Erlang Training & Consulting) wrote:
> There used to be an application called EVA, an Event and Alarm Handler, 
> which is discontinued and not supported any more. I didn't like it as it 
> was too tightly connected to SNMP. I assume others felt the same, as I 
> rarely saw others use it.
> 
> Francesco
> 
> 
> Serge Aleynikov wrote:
> 
>> Would it be correct to assume then that barely anyone uses the SASL's 
>> alarm_hangler?  If so, what's the reason for leaving it in the OTP, or 
>> otherwise, why don't make the basic functionality a bit more usable 
>> within the SASL framework?
>>
>> Serge
>>
>> Francesco Cesarini (Erlang Training & Consulting) wrote:
>>
>>> The SASL Alarm handler is a very simple handler, provided to allow 
>>> users to get started quickly. Somewhere in the documentation (And in 
>>> both our and Ericsson's OTP courses) users are encouraged to develop 
>>> their own handler and replace it with a more complex one.
>>>
>>> Regards,
>>> Francesco
>>> -- 
>>> http://www.erlang-consulting.com
>>>
>>> Serge Aleynikov wrote:
>>>
>>>> Is there a reason that the SASL's alarm handler doesn't check for 
>>>> duplicate alarms before adding new ones to the list?
>>>>
>>>> If you do the following on the system with low disk space / free 
>>>> memory:
>>>>
>>>> 1. Start SASL
>>>> 2. Start OS_MON
>>>> 3. Stop  OS_MON
>>>> 4. Start OS_MON
>>>>
>>>> you'll get duplicate alarms that are not easily clearled (you'd need 
>>>> to call alarm_handler:clear_alarm() multiple times for the same 
>>>> AlarmID...
>>>>
>>>> 17> alarm_handler:get_alarms().
>>>> [{system_memory_high_watermark,[]},
>>>>  {{disk_almost_full,"/mnt/cdrom"},[]},
>>>>  {{disk_almost_full,"/usr"},[]},
>>>>  {system_memory_high_watermark,[]},
>>>>  {{disk_almost_full,"/mnt/cdrom"},[]},
>>>>  {{disk_almost_full,"/usr"},[]}]
>>>> 18> alarm_handler:clear_alarm(system_memory_high_watermark).
>>>> [{{disk_almost_full,"/mnt/cdrom"},[]},
>>>>  {{disk_almost_full,"/usr"},[]},
>>>>  {system_memory_high_watermark,[]},
>>>>  {{disk_almost_full,"/mnt/cdrom"},[]},
>>>>  {{disk_almost_full,"/usr"},[]}]
>>>>
>>>> This looks like a bug to me.
>>>>
>>>> Serge