Do you use Android Support? I encountered a similar phenomenon and it turned out that the processes d.process.media and sdcard (as seen with top in the terminal) caused the battery drain, see this thread
I had a similar problem. This occurred after an attempt to transfer the file via BT. There was probably a transmission error. Helped restart the BT service. More details here
https://forum.sailfishos.org/t/systemd-journald-cpu-consumption/5140
Thanks for the recommendation, I will try it later.
However, am I wrong or SystemDataScope is similar to SysMon and only logs/displays the status of the CPU/Battery (among others), without logging which specific process was using a specific amount of CPU at a specific time?
Yes, I am using android support. I will take a look. Thanks!!
I never use BT, so I would discard it in my case.
Thanks, however! 
SystemDataScope can log data while the app is not running. It’s basically a frontend for the linux datalogger “collectd”.
To do so go to settings and enable the first two checkboxes (and the third if you want to start loggin at bootup)
Indeed, collectd will log overall stats. Not sure whether newer collectd would allow you to split stats between apps. I would have to update SDS and collectd to 4.0 new requirements (loss of statefs) and graph generation script to python3. Was not needed before, now it is time, it seems.
As for original issue, follow also CPU sleep. On phone, it is important to keep it sleeping to avoid battery drain. Although, case as yours, seems to be extreme.
I have the impression that there is some kind of process or “sequence of processes” that trigger the battery draining at some point.
As I said, after restarting, the phenomena disappears, and then it reappears after a couple of days.
After restarting the phone today in the morning, and with low-to-moderate use along the day, the battery has only lost around 15%-20% of charge.
I will keep investigating 
EDIT: this information is obsolete now, see comment below.
As has been mentioned, SystemDataScope is a frontend for collectd.
So it is possible to add configuration to monitor resource usage for a specific app by adding a configuration file to collectd:
# /etc/collectd.conf add the line:
Include /etc/collectd.d/*conf
And place something like this in /etc/collectd.d/monitor-processes.conf:
LoadPlugin exec
<Plugin exec>
#Exec "defaultuser" "/usr/libexec/collectd/monitor-processes.sh"
Exec "nemo" "/usr/libexec/collectd/monitor-processes.sh"
</Plugin>
The script can do something like this: monitor-processes.sh (note that this assumes non-busybox bash and ps)
#!/usr/bin/env bash
HOSTNAME="${COLLECTD_HOSTNAME:-localhost}"
INTERVAL="${COLLECTD_INTERVAL:-5}" # collectd will set this in its environment, 5 secs default only for calling manually
# suspect process names here:
processes="lipstick connmand sensorfwd msyncd sensors.qcom voicecall-ui"
gaugevalues="cpu mem"
absvalues="rss vsize"
for p in $processes; do
for v in $gaugevalues; do
VALUE=$(ps --no-headers -p $(/sbin/pidof ${p}) -o %${v} 2>/dev/null || echo "U")
LANG=C echo "PUTVAL \"$HOSTNAME/single-process/gauge-${p}_${v}\" interval=$INTERVAL N:$(echo $VALUE)"
done
for v in $absvalues; do
VALUE=$(ps --no-headers -p $(/sbin/pidof ${p}) -o ${v} 2>/dev/null || echo 0)
LANG=C echo "PUTVAL \"$HOSTNAME/single-process/absolute-${p}_${v}\" interval=$INTERVAL N:$(echo $VALUE)"
done
done
Now, this will just collect the data, It’s a bit more involved to then have that data graphed by the application. Haven’t done that part 
2 Likes
I have just released collectd and SystemDataScope which should work now on SFOS 4.0.1. If you want to follow some specific processes, take a look into Plugin processes. Default config (/etc/collectd.conf) has examples on how to enable monitoring some processes. You would have to enable plugin (uncomment corresponding LoadPlugin in the top of config), uncomment and specify your processes in plugin config (section as shown below)
<Plugin processes>
Process "collectd"
Process "rrdtool"
Process "systemdatascope"
Process "sensorfwd"
Process "sensors.qcom"
</Plugin>
and restart collectd. After running for some time, just regenerate graphs in SystemDataScope settings. Plotting processes stats is supported already (if it is broken, please open an issue).
8 Likes
That’s very cool, thanks for the update!
Thanks a lot!!! I will try it
This may sound silly, but if you can catch your device while it’s discharging rapidly then here is something else you could check.
Try logging the discharge rate while you toggle flightmode on and off a couple of times. Say at five-minute intervals.
I have seen power consumption rise sharply in flightmode, then fall back when I enable radios again. Once this has happened, it would keep doing it until next reboot. Never saw it immediately following a reboot, so my workaround is to avoid flightmode or to reboot after I enable it…
(I never could nail which process was the culprit, but I strongly suspect Email or CalDAV sync. Some race condition or incorrectly handled network error maybe.)
1 Like
sailsync-owncloud-daemon shows this behaviour sometimes.
Do you have that installed and enabled?
That’s interesting, but so far I haven’t set up Owncloud sync.
However, I plan to do it, so this is good to know.
Just to clarify, with Owncloud I mean the SailSync tool from OpenRepos, not Sailfish native NextCloud support.
Just a hint for those that try moitoring processes: it appears the “Process” directive does not pick up those processes (I suspect) whose names appear multiple times in the process list. This affects e.g. GUI apps which are launched through sailfish-qml or invoker.
For those, use ProcessMatch instead.
Example ps output:
nemo 2830 7111 0 08:17 ? 00:00:07 /usr/bin/sailfish-browser
nemo 7111 6486 0 Apr23 ? 00:00:03 /usr/libexec/mapplauncherd/booster-browser --systemd
nemo 14568 6486 0 09:15 ? 00:00:00 invoker -s --type=browser -G /usr/bin/sailfish-browser
nemo 14573 7111 11 09:15 ? 00:00:01 booster [browser]
So this won’t work:
Process "sailfish-browser"
but this does:
ProcessMatch "sailfish-browser" "sailfish-browser"
The second parameter is an regular expression to specify the match. Other examples:
ProcessMatch "android.hw" "android\.hardware\."
ProcessMatch "fpd" "(sailfish|encsfa)-fpd"
2 Likes
Hi, did you ever work out what was causing this? I seem to have similar behaviour, but didn’t spot this post in my search (didn’t look for overnight).
I noticed something similar before with mine
1 Like
Note/Update:
As #12 has been accepted/merged a while ago, you don’t need to edit the main configuration file any more in collectd 5.12.0.sfos.5 and later.
You can just add a “drop-in” file in /etc/collectd.d/foo.conf with your configuration parts. See /etc/collectd.d/processes.conf.example.