Closed jurriaan closed 5 years ago
Related https://github.com/dresden-elektronik/deconz-rest-plugin/issues/1031#issuecomment-447921594
@jurriaan Can you please send the zll.db file to mpi@dresden-elektronik.de so I can check whats wrong here.
@manup done!
Thanks, I've took the 2.05.51 offline for now until the issue is fixed.
https://github.com/dresden-elektronik/deconz-rest-plugin/issues/1050
Checked the zll.db it looks fine. Found the issue I'll do some more checks and then upload 2.05.52 as bugfix release later on.
Thanks a lot :)
@manup I just downgraded to 2.05.50, and it looks like my configuration is broken now. Is that to be expected? The names of all devices are missing and Phoscon is showing the commissioning screen.
Absolutely not :/ in the file you've just send me all is looking normal with device names and the auth apikeys.
Does it work when you close deCONZ and overwrite zll.db with the one from the mail, and open deCONZ again?
Also check that deCONZ is really closed, I needed to kill it the hard way
killall -9 deCONZ
@manup Yeah, had to do that as well. Not sure what went wrong, but after restoring the zll.db from the mail it works again.
Seeing the same (on my test system). deCONZ hangs on startup, consuming 100% CPU. The GUI isn't even started. It only shows the first two messages in the log:
$ sudo systemctl status deconz-gui
● deconz-gui.service - deCONZ: ZigBee gateway -- GUI/REST API
Loaded: loaded (/lib/systemd/system/deconz-gui.service; enabled; vendor preset: enabled)
Drop-In: /etc/systemd/system/deconz-gui.service.d
└─override.conf
Active: active (running) since Mon 2018-12-17 19:13:16 CET; 7min ago
Main PID: 27336 (deCONZ)
CGroup: /system.slice/deconz-gui.service
├─27336 /usr/bin/deCONZ --http-port=80 --dbg-info=1 --dbg-aps=2 --dbg-error=1
├─27346 dbus-launch --autolaunch 36338e1dca924dbd9a2413fc1346afd5 --binary-syntax --close-stderr
└─27347 /usr/bin/dbus-daemon --fork --print-pid 5 --print-address 7 --session
Dec 17 19:13:16 pi1 systemd[1]: Started deCONZ: ZigBee gateway -- GUI/REST API.
Dec 17 19:13:17 pi1 deCONZ[27336]: libEGL warning: DRI2: failed to authenticate
Dec 17 19:13:17 pi1 deCONZ[27336]: libpng warning: iCCP: known incorrect sRGB profile
Reverted back to v2.05.50, but deCONZ still hangs, even after kill -9
and resetting the RaspBee. I'm not sure if I ran the self-compiled REST API plugin with the database cleanup on this system. Database seems normal on quick inspection using sqlitebrowser.
Seeing the same (on my test system). deCONZ hangs on startup, consuming 100% CPU. The GUI isn't even started. It only shows the first two messages in the log:
I've messed with a NULL pointer... the pointer won :(
Are all deCONZ instances closed (and everything holding a lock on the database)?
ps ax | grep deCONZ
2.05.52 is online now to mitigate the issues
https://github.com/dresden-elektronik/deconz-rest-plugin/releases/tag/V2_05_52
Thou shalt not follow the NULL pointer, for chaos and madness await thee at its end. -- The 10 commandments for C programmers.
2.05.52 is online now
That version starts normally. Thanks, Manuel!
I actually made a backup before updating, but my production system starts OK using v2.05.52.
There was a 20-second delay during startup. Upon inspection, the device_descriptor
table again contained duplicate rows for the faulty lumi.ctrl_ln2.aq1
switch. I purged the duplicates, and now there's no more delay during startup.
I've disabled the automatic purge in 2.05.52, I'm not sure yet if it causes some heavy side effects (albeit I don't know how), since only few installations seem to have the issue it might be more safe to fix them manually.
Wim had a really bad start with 2.05.51 https://github.com/dresden-elektronik/deconz-rest-plugin/issues/1050
I purged the duplicates, and now there's no more delay during startup.
The code to prevent new duplicates is active, so in theory the database should stay clean now.
@manup 2.05.52 works fine, thanks for the quick fix!
I'm currently on 2.09.03-raspbian-buster-stable, and am not getting the above described behaviour ....
It keeps hanging in a 'DB cleanup' state while using 100% CPU.