[OCTAVIA] How to monitor amphora

Gaël THEROND gael.therond at gmail.com
Mon Nov 26 06:45:27 UTC 2018


I’ve taken some time to read your presentation and I have to say thank you
for your wonderful work ! It’s pretty much complete and with code example
which is really cool.

It’s exactly what I was looking for.
If I could just suggest one thing, it would be to give operators freedom to
choose which logging and which metrics clients they want to use.

Doing so would avoid having to rely on gnocchi as you would be able to
choose the clients type at amphora build time and specify clients target at
runtime using the amphora-agent.

On another hand, it might we’ll be tricky to do so as everything is
namespaced into the amphora instance.

As suggested using gnocchi to store HAProxy logs may well be the easiest
short path actually, I’m wondering if it would be possible for searchlight
to use that logs at some point.

Anyway, thanks for the awesome job.
For now I’ll keep it simple and just use the API for basic status.

Kind regards,

Le lun. 26 nov. 2018 à 04:23, Sa Pham <saphi070 at gmail.com> a écrit :

> yes. I have discussed with the Octavia team. PTL Johnson has suggested me
> to use gnocchi to store octavia amphora metric.
> On Wed, Nov 21, 2018 at 3:16 PM Gaël THEROND <gael.therond at gmail.com>
> wrote:
>> Hi! Thanks for this material, I’ll have a look at it.
>> As our philosophy at work is to always push upstream all our
>> patchs/features or bug fixes, I won’t modify and keep the source code on
>> our own and if we need further features like that I think we will rather
>> push for a Blueprint with all required commits.
>> Did you already discussed that topic with the Octavia team?
>> Thanks a lot.
>> Le mer. 21 nov. 2018 à 09:12, Sa Pham <saphi070 at gmail.com> a écrit :
>>> Hi,
>>> In Vietnam Openinfra Days 2018, I have a presentation about monitoring
>>> and logging for Octavia Amphora.  We have to customize octavia source code
>>> to do this. I send you my presentation slide:
>>> https://drive.google.com/file/d/1dHXExEKrHDg4Cf3D1fBeulLDW_G-txLr/view?usp=sharing
>>> Best,
>>> On Wed, Nov 21, 2018 at 3:05 PM Gaël THEROND <gael.therond at gmail.com>
>>> wrote:
>>>> Hi guys,
>>>> As already discussed I had to test Octavia as our corporate
>>>> LoadBalancer solution and it was a success.
>>>> Thank to everyone on this list that assisted me and especially Michael
>>>> Octavia is fully working and without weird nightmare glitches.
>>>> Now that I validated the technical part of my project, I need to enter
>>>> more operationals questions such as what’s the best way to monitor and log
>>>> amphora?
>>>> I would like to be able to get a monitoring and logging agent installed
>>>> on the amphora in order to get proper metrics about what’s going on with
>>>> the LoadBalancer, is it fully cpu loaded? Is it using all network resources
>>>> available, are the file descriptors near the initial limits? How much xxx
>>>> HTTP return code the loadBalancer is facing and send those logs to an ELK
>>>> or something similar.
>>>> Do you know if that something that I could achieve by adding more
>>>> elements to the image at DIB time or are there any other better
>>>> best-practices that I should be aware of ?
>>>> As Octavia is creating a namespace for each HAProxy process, I am
>>>> wandering if it’s even possible.
>>>> Thanks a lot for your hard work guys.
>>>> Kind regards,
>>> --
>>> Sa Pham Dang
>>> Cloud RnD Team - VCCloud
>>> Phone/Telegram: 0986.849.582
>>> Skype: great_bn
> --
> Sa Pham Dang
> Cloud RnD Team - VCCloud
> Phone/Telegram: 0986.849.582
> Skype: great_bn
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://lists.openstack.org/pipermail/openstack-discuss/attachments/20181126/409f7e6e/attachment-0001.html>

More information about the openstack-discuss mailing list