jimklimov
commented
1 month ago Interesting bits from a debug run:
- Original
ups.confcontents with comments, indentations, blank lines, etc.:# cat ups.conf #nowait #debug_min = 6 debug_min = 0
[dummy] driver = dummy-ups
debug_min = 6
port = dummy.dev[eco650] driver = "usbhid-ups" port = "auto" vendorid = "0463" productid = "FFFF" product = "Ellipse ECO" desc = "Eaton Ellipse ECO 650" serial = "000000000" vendor = "EATON" bus = "003"
pollonly
#debug_min = 0
debug_min = 2
* NDE artifacts (made a year ago and unchanged for way too long!):; cd /etc/systemd/system && grep . nut-driver@.service.d/
nut-driver@.service.d/nut-driver-enumerator-generated-checksum.conf:[Service] nut-driver@.service.d/nut-driver-enumerator-generated-checksum.conf:Environment='SECTION_CHECKSUM_GLOBAL=e66f236c1cb34ba08a795537493bc884'
nut-driver@dummy.service.d/nut-driver-enumerator-generated-devicename.conf:[Service] nut-driver@dummy.service.d/nut-driver-enumerator-generated-devicename.conf:Environment='DEVICE="dummy"'
nut-driver@eco650.service.d/nut-driver-enumerator-generated-checksum.conf:[Service] nut-driver@eco650.service.d/nut-driver-enumerator-generated-checksum.conf:Environment='SECTION_CHECKSUM=8a4a7aa5a28a7d38f1961f4d7c3913c9'
nut-driver@eco650.service.d/nut-driver-enumerator-generated-devicename.conf:[Service] nut-driver@eco650.service.d/nut-driver-enumerator-generated-devicename.conf:Environment='DEVICE="eco650"'
nut-driver@eco650.service.d/nut-driver-enumerator-generated.conf:# Customization generated Tue May 9 10:10:25 UTC 2023 by nut-driver-enumerator for NUT device 'eco650' nut-driver@eco650.service.d/nut-driver-enumerator-generated.conf:# DO NOT EDIT: This file would be removed or overwritten by that service nut-driver@eco650.service.d/nut-driver-enumerator-generated.conf:[Unit] nut-driver@eco650.service.d/nut-driver-enumerator-generated.conf:Description=Network UPS Tools - device driver for NUT device 'eco650' nut-driver@eco650.service.d/nut-driver-enumerator-generated.conf:Wants=systemd-udev.service systemd-udev-settle.service nut-driver@eco650.service.d/nut-driver-enumerator-generated.conf:After=systemd-udev.service systemd-udev-settle.service
:; ls -la nut-driver@.service.d/ -rw-r--r-- 1 root root 81 Apr 19 2023 nut-driver@.service.d/nut-driver-enumerator-generated-checksum.conf -rw-r--r-- 1 root root 39 Dec 1 15:32 nut-driver@dummy.service.d/nut-driver-enumerator-generated-devicename.conf -rw-r--r-- 1 root root 74 May 9 2023 nut-driver@eco650.service.d/nut-driver-enumerator-generated-checksum.conf -rw-r--r-- 1 root root 40 Dec 1 15:32 nut-driver@eco650.service.d/nut-driver-enumerator-generated-devicename.conf -rw-r--r-- 1 root root 364 May 9 2023 nut-driver@eco650.service.d/nut-driver-enumerator-generated.conf
* Sanitized config is slurped reasonably:- UPSCONF_DATA=debug_min=0 [dummy] driver=dummy-ups port=dummy.dev [eco650] driver=usbhid-ups port=auto vendorid=0463 productid=FFFF product="Ellipse ECO" desc="Eaton Ellipse ECO 650" serial=000000000 vendor=EATON bus=003 debug_min=2
- [ 218 != 0 ]
- SDP (Section-Driver-Port) highlight is determined well:
UPSCONF_DATA_SDP=[dummy] driver=dummy-ups port=dummy.dev [eco650] driver=usbhid-ups port=auto - [ 218 != 0 ]
- TO CHECK: Should "218" (char count of UPSCONF_DATA) be consulted in both places?
- List of devices in ups.conf is correct:
UPSLIST_FILE=dummy eco650 - [ 12 = 0 ]
- ...matches the known services:
UPSLIST_SVCS=dummy eco650 - [ 12 = 0 ]
- global section content found and checksummed:
- echo debug_min=0
- md5sum
- awk {print $1}
- _MD5=e66f236c1cb34ba08a795537493bc884 ...
- grep Environment='SECTION_CHECKSUM_GLOBAL= /etc/systemd/system/nut-driver@.service.d/nut-driver-enumerator-generated-checksum.conf
- sed -e s,^Environment='SECTION_CHECKSUM_GLOBAL=,, -e s,'$,,
- [ e66f236c1cb34ba08a795537493bc884 = e66f236c1cb34ba08a795537493bc884 ]
- for
dummydriver section, no checksum was saved before (and is not saved when the script re-runs): - echo [dummy] driver=dummy-ups port=dummy.dev
- md5sum
- awk {print $1}
- _MD5=ee8819cc0e7231ae99a05e3b7ec54618 ...
- [ -s /etc/systemd/system/nut-driver@dummy.service.d/nut-driver-enumerator-generated-checksum.conf ]
- echo Did not find '/etc/systemd/system/nut-driver@dummy.service.d/nut-driver-enumerator-generated-checksum.conf' with a SECTION_CHECKSUM
- return 1
- [ ee8819cc0e7231ae99a05e3b7ec54618 = Did not find '/etc/systemd/system/nut-driver@dummy.service.d/nut-driver-enumerator-generated-checksum.conf' with a SECTION_CHECKSUM ]
- [ 0 = 0 ]
- _TMP_SVC=dummy
** TO THINK: Is this related to the section having no config items beside SDP? ** TO THINK: Is it right to compare checksum to error message? eco650section checksum is detected as not matching the saved value:- echo [eco650] driver=usbhid-ups port=auto vendorid=0463 productid=FFFF product="Ellipse ECO" desc="Eaton Ellipse ECO 650" serial=000000000 vendor=EATON bus=003 debug_min=2
- md5sum
- awk {print $1}
- _MD5=6b073941959221404dc6397afd2e8190 ...
- grep Environment='SECTION_CHECKSUM= /etc/systemd/system/nut-driver@eco650.service.d/nut-driver-enumerator-generated-checksum.conf
- sed -e s,^Environment='SECTION_CHECKSUM=,, -e s,'$,,
- [ 6b073941959221404dc6397afd2e8190 = 8a4a7aa5a28a7d38f1961f4d7c3913c9 ]
- [ 5 = 0 ]
- _TMP_SVC=dummy eco650
- [ 12 = 0 ]
- Just after that: it does something that does not culminate in service reload nor saved new checksums even if it did conclude the correctly-looking NEW_CHECKSUM list:
... - echo dummy eco650
- return 1
- NEW_CHECKSUM=dummy eco650
- [ 0 != 0 ]
- [ 12 != 0 ]
- upslist_delSvcs
- common_isFiled dummy
- [ 12 != 0 ]
- [ dummy = dummy ]
- return 0
- common_isFiled eco650
- [ 12 != 0 ]
- [ eco650 = dummy ]
- systemd_validInstanceName dummy
- calc_md5 dummy
- echo dummy
- md5sum
- awk {print $1}
- _MD5=f02e326f800ee26f04df7961adbf7c0a
- [ -n f02e326f800ee26f04df7961adbf7c0a ]
- echo f02e326f800ee26f04df7961adbf7c0a
- echo MD5_f02e326f800ee26f04df7961adbf7c0a
- [ MD5_f02e326f800ee26f04df7961adbf7c0a = eco650 ]
- [ eco650 = eco650 ]
- return 0
- [ no = yes ]
- [ no = yes ]
- [ 12 != 0 ]
- upslist_addSvcs
- common_isRegistered dummy
- [ 12 != 0 ]
- [ dummy = dummy ]
- return 0
- common_isRegistered eco650
- [ 12 != 0 ]
- [ eco650 = dummy ]
- systemd_validInstanceName eco650
- calc_md5 eco650
- echo eco650
- md5sum
- awk {print $1}
- _MD5=725517ba9eb4f1b0435dff2aeff5e693
- [ -n 725517ba9eb4f1b0435dff2aeff5e693 ]
- echo 725517ba9eb4f1b0435dff2aeff5e693
- echo MD5_725517ba9eb4f1b0435dff2aeff5e693
- [ MD5_725517ba9eb4f1b0435dff2aeff5e693 = dummy ]
- [ eco650 = eco650 ]
- return 0
- systemd_refreshSupervizor
- /bin/systemctl daemon-reload
- upslist_readSvcs after checking for new config sections to define service instances
- systemd_listInstances
- systemd_listInstances_raw
- sed -e s/^.*@// -e s/.service$//
-
- sort -n /bin/systemctl show nut-driver@* -p Id
- grep -E =nut-driver
- sed s,^Id=,,
- UPSLIST_SVCS=dummy eco650
- [ 12 = 0 ]
- upslist_readSvcs
- systemd_listInstances
- systemd_listInstances_raw
- sed -e s/^.*@// -e s/.service$//
- /bin/systemctl show nut-driver@* -p Id
- sort -n
- grep -E =nut-driver
- sed s,^Id=,,
- UPSLIST_SVCS=dummy eco650
- [ 12 = 0 ]
- [ 12 != 0 ]
- echo === The currently defined service instances are: === The currently defined service instances are:
- echo dummy eco650 dummy eco650
- [ 12 != 0 ]
- echo === The currently defined configurations in '/etc/nut/ups.conf' are: === The currently defined configurations in '/etc/nut/ups.conf' are:
- echo dummy eco650 dummy eco650
- [ yes = yes ]
- systemd_restart_upsd
- /bin/systemctl is-active nut-server
- echo Reloading or restarting NUT data server to make sure it knows new configuration... Reloading or restarting NUT data server to make sure it knows new configuration...
- timeout 90s /bin/systemctl reload-or-restart nut-server
- UPSLIST_EQ_RES=0
- upslist_equals dummy eco650 dummy eco650
- [ 12 != 0 -a 12 = 0 ]
- [ 12 = 0 -a 12 != 0 ]
- [ dummy eco650 = dummy eco650 ]
- return 0
- calc_md5_file /etc/nut/ups.conf
- [ -s /etc/nut/ups.conf ]
- md5sum
- awk {print $1}
- _MD5=329a0839f13478c06217857d140a31c1
- [ -n 329a0839f13478c06217857d140a31c1 ]
- echo 329a0839f13478c06217857d140a31c1
- UPSCONF_CHECKSUM_END=329a0839f13478c06217857d140a31c1
- [ 329a0839f13478c06217857d140a31c1 != 329a0839f13478c06217857d140a31c1 ]
- [ 0 = 0 ]
- date -u
- echo Thu Apr 18 20:18:40 UTC 2024 : OK: No more changes to reconcile between systemd service instances and device configurations in '/etc/nut/ups.conf' Thu Apr 18 20:18:40 UTC 2024 : OK: No more changes to reconcile between systemd service instances and device configurations in '/etc/nut/ups.conf'
- [ = no ]
- return 42
- exit 42
It was brought to my attention by @ericclappier-eaton et al, that NDE-driven reload-or-restart of
nut-driver@units did not work for them after upgrading to NUT v2.8.2 release (as packaged in the Debian-derived OS of choice, I believe).This may potentially be related to work on #682 merged before the release, and if so - that would be a regression in this release. Possibly something broke even earlier, in 2.8.1.
I was able to reproduce the behavior on a development Proxmox system with current NUT master: editing or
touching/etc/nut/ups.conf(with the NDE path units active and started) does trigger a call to the script and a reload of thenut-serverunit; however it seems to do nothing about the driver units even when section content is edited (e.g.debug_minvalue changes):Similarly with command-line execution of the
nut-driver-enumerator.shscript which is more promising to debug the issue.According to script's log, it still does see and compare lists of devices in
ups.confand configured as systemd unit instances, so this part works. And then it decides that there areNo more changes to reconcile between systemd service instances and device configurations in '/etc/nut/ups.conf'which is the problem.