NEW Monitoring graph

maverick_slo

Yes but look at the graph and bubbles.
IMO it is showing me wrong data.

It should show 25+ not between 15-20…

maverick_slo

OMG sorry, sorry.
I feel like a dumbass :-[
All clear fine and dandy :)

peter808

@jdillard: do you need any further info?

peter808

Just tested it with an upgrade install from 2.2 to 2.3 beta: everything under "Monitoring" works, so it just seems to be a problem with the fresh install.

Maybe that helps.

jdillard

@peter808:

Just tested it with an upgrade install from 2.2 to 2.3 beta: everything under "Monitoring" works, so it just seems to be a problem with the fresh install.

Maybe that helps.

Thanks Peter. I saw it was an amd64 image, but which platform and type did you use to install? or the full name of the file would work.

peter808

pfSense-2.3-BETA-amd64-20160313-1302.iso

NOYB

Quick links pull request submitted.
https://github.com/pfsense/FreeBSD-ports/pull/86

Quick Links Details:
https://forum.pfsense.org/index.php?topic=107790.90

Branch:
Status_/Monitoring-Quick_Links
https://github.com/NOYB/FreeBSD-ports/tree/Status/Monitoring-_Quick_Links

Commit:
Status / Monitoring - Quick Links
Genesis
https://github.com/NOYB/FreeBSD-ports/commit/7f55b614a61ed100b6967c2da05a1255ab822e1f

Harvy66

@jdillard:

@maverick_slo:

@jdillard:

@maverick_slo:

Hello!

Im on latest snap and bubbles doesnt seem to have right values.
See attached.

BR,
G

Make sure you are looking at the correct axis

Ummm green bubble should have 25+ ms as value not below. Or am I missing something?

Green says "(right axis)" and is between the 15 and 20 ticks on the right axis

Just a random idea, but if two series are "similar" in scale, maybe lock them both to the same scale? Would help for these situations and allow a visual apples-to-apples comparison for these situations. Of course if two series are too dissimilar, they would need to be different scales.

jdillard

I've thought about adding a "match axis scales" as a manual option.

(it's been in the code as a TODO) https://github.com/pfsense/FreeBSD-ports/blob/1f07533cd85299ceee0b44115b201f64969d8085/sysutils/pfSense-Status_Monitoring/files/usr/local/www/status_monitoring.php#L1584

jdillard

@NOYB:

Quick links pull request submitted.
https://github.com/pfsense/FreeBSD-ports/pull/86

Quick Links Details:
https://forum.pfsense.org/index.php?topic=107790.90

Branch:
Status_/Monitoring-Quick_Links
https://github.com/NOYB/FreeBSD-ports/tree/Status/Monitoring-_Quick_Links

Commit:
Status / Monitoring - Quick Links
Genesis
https://github.com/NOYB/FreeBSD-ports/commit/7f55b614a61ed100b6967c2da05a1255ab822e1f

Thanks NOYB, I've been looking it over, I just want to make sure it is done right since it touches the config.

jahonix

@jdillard:

I've thought about adding a "match axis scales" as a manual option.

That would be awesome!
I had screenshots where you couldn't determine which horizontal line is which - but didn't want to interfere with other stuff you're currently working on so I didn't post that.

While I'm at it: if you deselect some of the graphs shown and hit "update" then your selection is reset. Can you store which graph is on or off?

jdillard

@jahonix:

@jdillard:

I've thought about adding a "match axis scales" as a manual option.

That would be awesome!
I had screenshots where you couldn't determine which horizontal line is which - but didn't want to interfere with other stuff you're currently working on so I didn't post that.

While I'm at it: if you deselect some of the graphs shown and hit "update" then your selection is reset. Can you store which graph is on or off?

These features are drifting into "nice to haves" territory so, while they are doable, they are less of a priority. I'll try to open a TODO ticket as a reminder.

The "remembering which lines you have selected" part is going to be difficult in comparison, I wouldn't expect that anytime soon. It should be a fun problem to solve though.

jdillard

Another idea I had (and you may want to play around with it NOYB) is to have a monitoring dashboard widget that had basically the same config options as the monitoring page (after the auto update is included) and you would be able to add multiple monitoring widgets that can be configured differently. There is some custom work to the widgets that would need to be done (without breaking the current widgets) to allow this, but it would be pretty cool.

A first step would be to have a monitoring widget.

maverick_slo

Just a random idea…
What if we could have multiple dashboards :)?
Like first page but multiple and different widgets on each page.

peter808

@jdillard:

Another idea I had (and you may want to play around with it NOYB) is to have a monitoring dashboard widget that had basically the same config options as the monitoring page (after the auto update is included) and you would be able to add multiple monitoring widgets that can be configured differently. There is some custom work to the widgets that would need to be done (without breaking the current widgets) to allow this, but it would be pretty cool.

A first step would be to have a monitoring widget.

That widget would be great for the future !

But maybe for the moment it would be enough (and easier for you!) to implement the option for multiple, adjustable graphs on the monitoring page. This would also give you a very deep first look on whats going on.

jahonix

@jdillard:

…(after the auto update is included)...

Will auto update reset the selected graphs as well?
(where is the selection done - probably directly in the browser, right?)

whitexp

@cmb:

@whitexp:

PHP Errors:
[10-Mar-2016 15:07:02 America/Sao_Paulo] PHP Fatal error: Call to undefined function rrd_fetch() in /usr/local/www/rrd_fetch_json.php on line 125

That means you don't have pecl-rrd, which has been a dependency since this was in place. What do you see for 'pkg info|grep pecl-rrd'? Should look something like:
[2.3-BETA][root@fw1]/root: pkg info|grep pecl-rrd
pecl-rrd-1.1.3_2               PHP bindings to rrd tool system

[2.3-BETA][admin@white.localdomain]/root: pkg info|grep pecl-rrd
pecl-rrd-1.1.3_2 PHP bindings to rrd tool system
[2.3-BETA][admin@white.localdomain]/root:

Crash report details:

PHP Errors:
[17-Mar-2016 09:41:09 America/Sao_Paulo] PHP Fatal error:  Call to undefined function rrd_fetch() in /usr/local/www/rrd_fetch_json.php on line 128
[17-Mar-2016 09:41:19 America/Sao_Paulo] PHP Fatal error:  Call to undefined function rrd_fetch() in /usr/local/www/rrd_fetch_json.php on line 128
[17-Mar-2016 09:45:56 America/Sao_Paulo] PHP Fatal error:  Call to undefined function rrd_fetch() in /usr/local/www/rrd_fetch_json.php on line 128
[17-Mar-2016 09:46:17 America/Sao_Paulo] PHP Fatal error:  Call to undefined function rrd_fetch() in /usr/local/www/rrd_fetch_json.php on line 128

Filename: /var/crash/minfree
2048

i still have the problem …

Interactive Graph
Error: SyntaxError: Unexpected token F

NOYB

@jahonix:

@jdillard:

…(after the auto update is included)...

Will auto update reset the selected graphs as well?
(where is the selection done - probably directly in the browser, right?)

If you mean will hidden series be restored, yes. It is no different than manually hitting the refresh button. It's just done for you at the selected a interval.

Well there is at least one difference. The loading graph banner isn't displayed.

MikeV7896

So I haven't looked at the monitoring graph since I updated last night (and I'm not at home to look now)… has the last sample being 0 been fixed on the graphs yet? A couple of days ago I noticed that all lines still returned to 0 on the right side of the graph.

NOYB

@virgiliomi:

So I haven't looked at the monitoring graph since I updated last night (and I'm not at home to look now)… has the last sample being 0 been fixed on the graphs yet? A couple of days ago I noticed that all lines still returned to 0 on the right side of the graph.

No.

Odds of having an RRD data point for the graph end time can be greatly increased though by updating/refreshing the graph near the end of the minute. Say like at hh:mm:59.

I committed a hack to the auto update pull request that syncs the refreshes to seconds 59 and it seems to work pretty well to prevent the graph from ending with values of zero (0). Once in a while it will if the RRD is updated between mm:59 and mm+1:00. Or the sync drifts off a bit and hits mm:00 once in a while.

But I think the real way to fix it is to use the most recent RRD data point time as the graph end time. And set graph start time to graph end time minus time period. Think something like that is how it works in previous version.