POLL: which is the feature we should implement next? #436

Open
ktsaou opened this Issue May 22, 2016 · 29 comments

Projects

None yet

10 participants

@ktsaou
Member
ktsaou commented May 22, 2016 edited

Just like the comments below

(I added one like to make the buttons available - please don't comment in this issue)

@ktsaou
Member
ktsaou commented May 22, 2016 edited

longer data history

find ways to compress, dedup, reduce or whatever necessary to have at least a week of data in netdata and possibly longer on-disk archived durations (monthly, annually, etc)

@ktsaou
Member
ktsaou commented May 22, 2016 edited

data snapshots

allow taking snapshots of time frames in full detail, that can be loaded back for further examination
#309

@ktsaou
Member
ktsaou commented May 22, 2016

multi-server dashboards

using the registry features #416 allow dynamic cross-server dashboards to be created and work towards making netdata a master piece for multi-server performance monitoring

@ktsaou
Member
ktsaou commented May 22, 2016 edited

health monitoring / alarms / alerts / notifications

add health monitoring features to netdata and create a central health monitoring system which will be used by all netdata installations to propagate their alarms

related issue: #318

@ktsaou
Member
ktsaou commented May 22, 2016

dashboard editor

allow creating custom dashboards directly from the web front end

@ktsaou ktsaou changed the title from poll: which is the feature we should implement next? to POLL: which is the feature we should implement next? May 22, 2016
@Yenthe666

Hi @ktsaou just my five cents here as a new person that just discovered Netdata (also congratz on this amazing software!):

  • It would be great to have the installation automatically creating an automatic daemon
  • The ability to create dashboards from the website is a major plus, this means no longer having to edit and create files from the command line.
  • Alarms and notifications would be awesome, especially if they could be sent to a push notification client for example.
@ktsaou
Member
ktsaou commented May 22, 2016

It would be great to have the installation automatically creating an automatic daemon

I would love to understand what this means for you. Can you open a new issue to describe it?

@Yenthe666

@ktsaou explained the raw details here #438.
Also: It would be great if data was stored in database such as postgreSQL.

@rvanrootselaar
  • historic data
@ktsaou
Member
ktsaou commented May 26, 2016 edited

historic data

This is what I thought too and this is why I wrote it first. However, as you can see most users need health monitoring before that. Let's see how it ends. I'll leave the poll running for a few more days...

@rvanrootselaar

Looking at it a little closer with a programmer colleague of mine we think that a pause button would even be better than historic data and probably be easier to realize. So just being able to keep the 1 hour view as it is without losing data past 1 hour ago (and not getting any new data). This means when a performance issue takes place and lasted, say a couple of minutes, you have more than one hour to analyse and look at the event before you lose all the data on it. Congrats on a really nice product by the way.

@ktsaou
Member
ktsaou commented May 28, 2016 edited

Looking at it a little closer with a programmer colleague of mine we think that a pause button would even be better than historic data and probably be easier to realize. So just being able to keep the 1 hour view as it is without losing data past 1 hour ago (and not getting any new data). This means when a performance issue takes place and lasted, say a couple of minutes, you have more than one hour to analyse and look at the event before you lose all the data on it.

Snapshots #309 is probably a better alternative to this. Using snapshots you don't pause data collection, but you save the timeframe you are interested. The saved snapshot could be loaded back to be analyzed and will be kept until you delete it.

Snapshots can also have a "killer feature": compare the current data against the snapshot data and present the differences, sorted by max difference. This way you could quickly find out that your disk is getting twice as many IOPs, your database is having twice as many slow queries, etc.

@HuangShaoyan

Allow using apt-get to install netdata on Debian and Ubuntu.

@fnzv
fnzv commented Jun 4, 2016

Storing data for retention on a backend like Elasticsearch, so it's also possible to send alerts & full text search

@jasonsoul

Hi @ktsaou ,
I have to say it's a great project it work perfectly with my Synology NAS.
Just wondering to know if there's any plan to monitor disk temperature?
Thanks

@ktsaou
Member
ktsaou commented Jun 7, 2016

Just wondering to know if there's any plan to monitor disk temperature?

Well, it can be easily done. I resisted to the idea since a few disks lead to corruption if you access SMART while they are writing. You can write a plugin in BASH for that. Check the charts.d directory. Pick an example chart and modify it to call hddtemp. If you need help, open a new issue and I'll help.

@nixfu
nixfu commented Jun 7, 2016

Move away from using the build scripts as installer. Build binaries and install packages, then use the install packages for the installation not the build scripts.

This was referenced Jun 10, 2016
@paulfantom
Contributor

@jasonsoul monitoring disk temperatures is already implemented (#511).
@nixfu @HuangShaoyan I am working on using OBS as a build and distribution platform.

@tx7
tx7 commented Jun 22, 2016 edited

Hi,
Can you add CPU temperature to the system overview, and its separated section?

@paulfantom
Contributor

@tx7 aren't CPU temperatures monitored by sensors.chart.sh plugin?

@tx7
tx7 commented Jun 22, 2016

I'm not sure.

@paulfantom
Contributor
paulfantom commented Jun 22, 2016 edited

@tx7 Documentation says they are.

@tx7
tx7 commented Jun 22, 2016 edited

So, I need to modify the codes in /etc/netdata/sensors.conf to enable it ?

# the directory the kernel keeps sensor data
sensors_sys_dir="${NETDATA_HOST_PREFIX}/sys/devices"

# how deep in the tree to check for sensor data
sensors_sys_depth=10

# if set to 1, the script will overwrite internal
# script functions with code generated ones
# leave to 1, is faster
sensors_source_update=1

# how frequently to collect sensor data
# the default is to collect it at every iteration of charts.d
sensors_update_every=

# array of sensors which are excluded
# the default is to include all
sensors_excluded=()
@paulfantom
Contributor

@tx7 It should work out-of-the-box. If it is not working add sensors=yes to /etc/netdata/charts.d.conf.
When it still doesn't work, maybe you have some strange sensors, check if sensors command gives you some output.
And if it still doesn't work, open an issue and we will try to resolve it.

@ktsaou
Member
ktsaou commented Jun 22, 2016

netdata finds sensors with this:

find /sys/devices -maxdepth 10 -name \*_input -o -name temp

on my laptop it gives this:

# find /sys/devices -maxdepth 10 -name \*_input -o -name temp
/sys/devices/virtual/hwmon/hwmon0/temp1_input
/sys/devices/virtual/thermal/thermal_zone0/temp
/sys/devices/virtual/thermal/thermal_zone1/temp
/sys/devices/platform/coretemp.0/hwmon/hwmon1/temp3_input
/sys/devices/platform/coretemp.0/hwmon/hwmon1/temp1_input
/sys/devices/platform/coretemp.0/hwmon/hwmon1/temp2_input
@tx7
tx7 commented Jun 23, 2016

I checked, the sensors are working.
I'm thinking to put a CPU temp in the system overview as quick glance.

@ktsaou
Member
ktsaou commented Aug 15, 2016

As requested, I have developed health monitoring!

It is in testing phase.

If you are interested to join the beta testers, join the discussion on PR #775

@shadycuz
Contributor
shadycuz commented Nov 3, 2016

@ktsaou Where are we at on this? I'm currently looking through things with the your help needed label trying to stay busy. Should you close and reopen this topic for another round of feature voting? I think health alarms are out of "Beta" testing but I know they are currently being expanded and improved.

@ktsaou
Member
ktsaou commented Nov 3, 2016

Alarms are still missing a few key features:

  1. central notifications, which will also provide alarms/notifications in a case a server crashes and provide a few new notification methods, like centralised web browser notifications for servers that their dashboard is not open. It will also provide the basics for having a chat bot interacting with all our netdata installs.
  2. a better UI at the dashboard.
  3. ability to setup and dismiss alarms from the dashboard, which however depends on authentication.

I think however it is time to work on the dashboard a bit more. I would like to move all documentation at the netdata dashboard and at the same time implement a few features related to the second winner of the poll: multi-server dashboards.

Another solution could be to implement authentication now (make netdata an OAUTH2 provider, so that all our netdata installs will use that one for authentication).

I don't know yet. My schedule is very tight, so I think I will just provide small improvements for this month.

Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment