Bug 12890 - ACPI Thermal-Zone Temp Graph - temp9A0 is not a vname nor a number
Summary: ACPI Thermal-Zone Temp Graph - temp9A0 is not a vname nor a number
Status: CLOSED FIXED
Alias: None
Product: IPFire
Classification: Unclassified
Component: --- (show other bugs)
Version: 2
Hardware: x86_64 Unspecified
: - Unknown - Minor Usability
Assignee: Leo Hofmann
QA Contact:
URL:
Keywords:
Depends on:
Blocks:
 
Reported: 2022-06-28 16:21 UTC by IPcopMann
Modified: 2023-04-18 16:37 UTC (History)
4 users (show)

See Also:


Attachments
Screen grab Status > Hardware Graphs (35.39 KB, image/jpeg)
2022-06-28 16:21 UTC, IPcopMann
Details

Note You need to log in before you can comment on or make changes to this bug.
Description IPcopMann 2022-06-28 16:21:11 UTC
Created attachment 1060 [details]
Screen grab Status > Hardware Graphs

Menu Status > Hardware Graphs been giving this error for quite a few releases.
I am on CU169 testing now, error still there.
Only get the error on a couple of boxes of identical hardware (Intel based fanless mini-PC of Chinese origin).

https://fireinfo.ipfire.org/profile/9e370b2f991acf24f3fc4df21bdee5ec15f9d622
Comment 1 Michael Tremer 2022-06-29 18:03:00 UTC
@Leo: Is this something for you?
Comment 2 Leo Hofmann 2022-07-10 10:58:59 UTC
My first guess would be that the board is reporting faulty or non-existent sensors?

The ACPI thermal-zone data should be logged in this directory:
/var/log/rrd/collectd/localhost/thermal-thermal_zone*

Could you please check if there is meaningful RRD data there?

I think this is the collectd plugin with which the data is logged:
https://collectd.org/wiki/index.php/Plugin:thermal
Comment 3 siosios 2022-07-20 01:15:41 UTC
This is also happening on cpu frequency on my equiptment, see my post below

https://community.ipfire.org/t/cpufreq-graph-error/8246
Comment 4 Leo Hofmann 2022-07-21 13:40:48 UTC
Hi siosios, thanks for figuring out the problem with the colors. I would have never thought of that. I'll be able to start working on this soon :)
Comment 5 IPcopMann 2022-07-22 12:38:24 UTC
I used rrddump to look at data files and all had just NaN entries (i.e. sensor failure).

Followed instructions to find new sensor hardware:
https://wiki.ipfire.org/installation/hardware-change

Newly recreated files still have NaN entries back to 2021, here is a sample extract:


    <!-- 2022-07-18 08:29:00 BST / 1658129340 --> <row><v>NaN</v></row>
                        <!-- 2022-07-18 15:48:00 BST / 1658155680 --> <row><v>NaN</v></row>
                        <!-- 2022-07-18 23:07:00 BST / 1658182020 --> <row><v>NaN</v></row>
                        <!-- 2022-07-19 06:26:00 BST / 1658208360 --> <row><v>NaN</v></row>
                        <!-- 2022-07-19 13:45:00 BST / 1658234700 --> <row><v>NaN</v></row>
                        <!-- 2022-07-19 21:04:00 BST / 1658261040 --> <row><v>NaN</v></row>
                        <!-- 2022-07-20 04:23:00 BST / 1658287380 --> <row><v>NaN</v></row>
                        <!-- 2022-07-20 11:42:00 BST / 1658313720 --> <row><v>NaN</v></row>
                        <!-- 2022-07-20 19:01:00 BST / 1658340060 --> <row><v>NaN</v></row>
                        <!-- 2022-07-21 02:20:00 BST / 1658366400 --> <row><v>2.0000000000e+01</v></row>
                        <!-- 2022-07-21 09:39:00 BST / 1658392740 --> <row><v>2.0000000000e+01</v></row>
                        <!-- 2022-07-21 16:58:00 BST / 1658419080 --> <row><v>2.0000000000e+01</v></row>

Fairly sure that's what's tripping up graphs.
Is there an easy way to delete old entries?
Comment 6 siosios 2022-08-02 15:42:51 UTC
I did a temporary fix for the cpufreq graph issue by limiting the number of cpu's it shows. I have updated my post listed above that shows what I've done.
Comment 7 siosios 2023-02-13 05:06:01 UTC
Updated my post as to what I’ve found. Sorry, I never learnt how to create patches otherwise I would submit it.
Comment 8 Bernhard Bitsch 2023-02-23 13:13:29 UTC
It is indeed just the selection of colors.
A code sequence

for i = 0..#cpus {
  j= i+1
  .... colors{'color1'.j} ...
}

produces legal indices into colors{} ( keys 'color1'...'color25' ) for i = 0..8 only.
Replacing j by i functions for #cpus<=25.
Whether the colors are suitable has to be checked.