[openstack-dev] [kolla] Heka v ELK stack logistics

Michal Rostecki mrostecki at mirantis.com
Fri Jan 15 07:39:53 UTC 2016


On 01/14/2016 03:56 PM, Michał Jastrzębski wrote:
> On 14 January 2016 at 04:46, Eric LEMOINE <elemoine at mirantis.com> wrote:
>> On Wed, Jan 13, 2016 at 1:15 PM, Steven Dake (stdake) <stdake at cisco.com> wrote:
>>> Hey folks,
>>>
>>> I'd like to have a mailing list discussion about logistics of the ELKSTACK
>>> solution that Alicja has sorted out vs the Heka implementation that Eric is
>>> proposing.
>>>
>>> My take on that is Eric wants to replace rsyslog and logstash with Heka.
>>
>>
>> See my other email on this point.  At this point, given the
>> requirements we have (get logs from services that only speak syslog
>> and write logs to local files), we cannot guarantee that Heka will
>> replace Rsyslog.  We are going to test the use of Heka's UdpInput
>> (with "net" set to "unixgram") et FileOutput plugins for that.  Stay
>> tuned!
>
> Yeah, also please try out different configs of only-rsyslog services.
> Maybe Mariadb can be set up in a way compliant with heka?
>

In case of MariaDB I see that you can

- do --skip-syslog, then all things go to "error log"
- set "error log" path by --log-error

So maybe setting stdout/stderr may work here?

I'd suggest to check the similar options in RabbitMQ and other 
non-OpenStack components.

I may help with this, because in my opinion, logging to stdout is the 
best default option for Mesos and Kubernetes - i.e. Mesos UI shows a 
"sandbox" of application, which generally is stdout/stderr. So if 
someone will not want to use Heka/ELK, then having everything in 
Mesos/Kubernetes would be probably the lesser evil than trying rsyslog 
to work here.

Cheers,
Michal



More information about the OpenStack-dev mailing list