[Linux-HA] Re: [Linux-ha-dev] If suicide is the answer, you're asking the wrong question

[Andrew Beekhof]

On Nov 8, 2007, at 11:40 PM, Lars Marowsky-Bree wrote:

> On 2007-11-08T17:31:04, Andrew Beekhof <beekhof at gmail.com> wrote:
>
>> My understanding of the proposal was that "any child process  
>> exiting, ever"
>> is counted as "so severe"
>> At least it would have to be that way to make it relevant to bug  
>> 1762.
>
> Probably.
>
> As we've discussed on IRC, one could invent a special exit code to  
> mean
> "this is not a fatal exit, just respawn me" (just like we have one  
> which
> says "do not respawn me" or during shutdown), but it seems a sensible
> default.
>
>> Assuming that a process exiting is a "not-yet-recoverable" error.
>> However for the mostly likely cases, at least in the crm, this is  
>> not true.
>
> Can you enlighten me on the most likely cases ...?

Basically any time CRM_ASSERT is called - which is also  
indistinguishable from a SEGV except by reading the logs or backtrace.
The CRM will specifically tell heartbeat if something happened from  
which it can never recover by using the "no respawn" exit code.