[sysmon] System Monitoring

Wallace said:
I noticed that it has shrunk, thank you for that. However, and this is probably just me, the icon still seems to 'jump out of the screen' too much. I know it's been mentioned before, but why not put it under the Diagnostics menu?
Click to expand...

It may move there eventually, but I'd need to extend the plugin API to support that. At the moment plugins can only place icons on the main screen or in the settings page.

As an aside, my comment about the green colour of the graphics in Sysmon were made before I realised that you could click on the chart and break the information down into bite sized chunks, so apologies for that.
Click to expand...

I'm happy to tweak the colours - if you want to take a look they're defined in /mod/var/mongoose/plugin/sysmon/temp.jim as far as I remember, just as a list of HTML colour codes.
 
I knew there was a reason, but couldn't remember it! Senior moment (again).

I might play with the graph colours some time, but as I say, I commented before I realised you could break things down and display them individually.

Everything is evolving at a pace, you blink your eyes and another option/button appears!
 
4291 said:
I have auto-update installed and my box is now on webif 0.9.7-2. I also have disable-ota installed, but the 04:30 Auto Update for Tuesday has reappeared in Scheduled Events.
Click to expand...
Just to update, the 04:30 Auto Update was eventually removed and hasn't reappeared.
 
OK, I've been holding my comments until I finally saw all four fan speeds on my gauge (including "off") - I'm not watching all the time and the full speed only occurs briefly, so I kept missing it despite the warnings from the sysmon graph. I am surprised the cooling curve is apparently so linear under the circumstances, but it all adds to experience.

1. If the fan is off it remains off until the upper temperature limit is reached.
2. If the fan is at any of its three "not off" throttle settings, the current speed is controlled by the temperature thresholds (ie if the temperature of the HDD rises again before it cools to the "fan off" threshold, the fan speeds up accordingly).
3. Even a very slight obstruction in the exhaust air flow has a significant effect on the cooling profile. From this I conclude that the ambient temperature will also have a significant effect. My ambient temp is 20 deg.C; if the room or the inside of a cabinet were much greater than this it is likely the fan would be on most or all of the time, possibly at full throttle and possibly accounting for some people's annoyance.
 
How do you account for the fact that the fan turns on with no new change in HDD temperature e.g.
12:35 to 12:40 55C Fan off (6 measurements)
12:41 55C Fan On Full
Also (in my case) the fan turns on some times during the 55C temerature range (but not at the begging) and sometimes during the 56C temerature range. Maybe HDD temperature + a time delay
 
I think it is likely we are not seeing exactly the same temperature reading as the Humax is making its decisions on. It is possible, for example, that the fan has its own regulator electronics and does not involve software at all. I wouldn't immediately jump to that conclusion though, the switching seems remarkably well correlated with the SMART data. Possibly the firmware uses a direct IO access to the sensor and gets an 8-bit value rather than a rounded integer.

I have not watched mine closely enough to see when exactly the switch points are relative to reported temperature, but the high trip seems to be 55 and the low point 50 (as reported)
 
Obvious question. Have you had it installed long enough to gather the data?


Sent from my iPhone 4s using Tapatalk
 
xyz321 : The jumps are probably due to the 1 minute sampling interval.
Click to expand...

Ezra Pound said:
But the 1 Minute intervals are 55, 55, 55, 55, 55, 55, 55, 54, 54, 54, 54, 51, 51, 51, 49, 49, 49, 49, 47, 47, 47, 47, 47, 47, 47
Click to expand...

Just a quick update, I've taken hundreds of HDD temperature measurements at 5 and 1 Second intervals and finally the penny dropped, the figures being read are only updated every 175 Seconds (about 3 Minutes) on the HDD itself, so you’ll never see a change more often than that
 
Interesting. If the fan speed gear change intervals are also in 175s quanta that is good evidence it is controlled by the SMART reading - if not, it must be coming from somewhere else.
 
I only recorded fan times to the nearest minute (more testing requiredOn) but it does seem to tie-up e.g. Fan Full on >> 6Mins. >> Half On>> 6Mins. >> Quarter >> 3Mins. >> Fan Off
 
I've done a little monitoring this morning and have reached the same (preliminary) conclusion. However, the fan transitions do not seem to be in sync with the sysmon graph.
 
af123 said:
I've just uploaded a new version of sysmon which uses a different graphing library. This one still renders graphs on the client side to avoid putting load on the Humax but should work on iOS and Android too - Android needs to download an additional library but that should happen automatically...
Click to expand...

Did some users have trouble with the old-style graphs on Android? They worked OK for me (HTC Desire HD), although the new ones render a lot quicker, so many thanks for the update!
 
Ezra Pound said:
Just a quick update, I've taken hundreds of HDD temperature measurements at 5 and 1 Second intervals and finally the penny dropped, the figures being read are only updated every 175 Seconds (about 3 Minutes) on the HDD itself, so you’ll never see a change more often than that
Click to expand...

Black Hole said:
Interesting. If the fan speed gear change intervals are also in 175s quanta that is good evidence it is controlled by the SMART reading - if not, it must be coming from somewhere else.
Click to expand...

Ezra Pound said:
I only recorded fan times to the nearest minute (more testing requiredOn) but it does seem to tie-up e.g. Fan Full on >> 6Mins. >> Half On>> 6Mins. >> Quarter >> 3Mins. >> Fan Off
Click to expand...

Black Hole said:
I've done a little monitoring this morning and have reached the same (preliminary) conclusion. However, the fan transitions do not seem to be in sync with the sysmon graph.
Click to expand...

I have some hard data, gathered by making a video of my gauge in operation and then timing the transitions using the playback index (accurate to the second). I have not so far correlated the timings with the sysmon graph.

Using the transition from off to full as the datum, transition to half occurred at 8m00s, transition to quarter at 12m00s, transition to off at 20m00s. This blows the 175s theroy.
 
The 175 Seconds is the rate at which the S.M.A.R.T. data is updated, the table below has 1 Second measurements, only Changes are logged, It's also looking like 175 really is 180, my script was loosing a few miliseconds each measurement

Code: 
Contents of /mod/tmp/temp_change.log 431.00 bytes
Log Temp. Change each Second
Sun May 13 18:00:58 BST 2012    055
Sun May 13 18:03:25 BST 2012    054
Sun May 13 18:06:24 BST 2012    053
Sun May 13 18:09:24 BST 2012    050
Sun May 13 18:12:24 BST 2012    048
Sun May 13 18:15:25 BST 2012    047
Sun May 13 18:18:25 BST 2012    046
Sun May 13 18:21:24 BST 2012    047
Sun May 13 18:27:24 BST 2012    048
Sun May 13 18:30:24 BST 2012    049
Sun May 13 18:36:24 BST 2012    050
Sun May 13 18:39:24 BST 2012    051
Sun May 13 18:45:23 BST 2012    052
Sun May 13 18:51:24 BST 2012    053
Sun May 13 18:57:24 BST 2012    054
Sun May 13 19:03:23 BST 2012    055
Sun May 13 19:12:24 BST 2012    053
Sun May 13 19:15:24 BST 2012    050
 
OK, here is the result of me correlating a gauge video with the sysmon chart:

FanSpeed1.png

Notice that fan and temperature change events seem to occur on 3 minute quanta, but my previous fan log had events at 4 minute quanta. I can only reconcile these by suggesting the underlying quantum is 1 minute. More data required!

Also notice that the fan is responding to events which do not show up in the SMART data.
 
I have never seen a temperature change time that was not a multiple of 3 Mins. - Yet e.g. 3 or 6 (possibly 9 12 etc) but never less than 3
 
You must log in or register to reply here.
Back
Top