On 09/02/15 14:40, Brian Stansberry wrote:
On 2/8/15 4:28 PM, James Livingston wrote:
> On Thu, 2015-02-05 at 09:00 +0000, Darran Lofthouse wrote:
>> Anyway the purpose of this thread is that I wanted to try and gather
>> together the kinds of warnings that we could be outputting, below is a
>> list of some I have thought of already but would be interested in
>> hearing any additional ideas.
>>
>> - SSL is not configured.
>> - SSL certificates are due to expire.
>> - Plain text password detected in the configuration.
>> - Some form of file based storage in use but growing beyond intended size.
>> - Default node name has not been changed.
>> - Patches available but not applied, subsequent releases available.
>
> If you're thinking about runtime warnings, as opposed to plain
> administrative actions that should be taken, there are some things which
> are WARN/ERROR in logs that may be good to show too so they don't get
> lost. For example timeouts acquiring pooled resources, such as EJBs
> instances or JDBC connections, which may indicate problems or
> mis-configuration.
>
> I have a half-baked idea about detecting excessive GC pauses in managed
> servers, which if it worked could make a useful runtime warning.
>
This is drifting into a general notification and alerting tool, which is
a natural direction to go.
Darran, do you have any thoughts on the boundaries between what you
envisioned and full fledged notification and/or alerting tool?
That is kind of what I am trying to explore with this thread. The types
of notifications I know I need are fairly simple for me to design
something around but I am trying to see the other kinds of notifications
we could consider so we don't design ourselves away from being able to
add them in the future.