- This topic has 47 replies, 5 voices, and was last updated May 21-5:04 pm by olsztyn.
-
Posts
-
SamK
This seems to happen regardless connection is WiFi or wired.
I cannot reproduce this behaviour with antiX-17.4.1-full kernel 4.9.160.
I can reproduce the behaviour with antiX-19.2-full as shippedConnectshares has not changed between v17.x and 19.x so is not the cause of the issue.
Something has changed in in antiX 19 which is causing the problem.
Initial suspect is something within the init.d system. In 2014 exactly the same symptoms and error message were seen for wifi only, not cable. It was traced to the sequence of tasks run in the shutdown and reboot sections of init.d and a fix put in place. That has worked OK upto antiX-19. At this point I have few spare resources to investigate what has changed. Perhaps another team member who has done work on the init system for antiX-19 might be able to get to it before me.
olsztyn
Something has changed in in antiX 19 which is causing the problem.
Initial suspect is something within the init.d system. In 2014 exactly the same symptoms and error message were seen for wifi only, not cable. It was traced to the sequence of tasks run in the shutdown and reboot sections of init.d and a fix put in place. That has worked OK upto antiX-19. At this point I have few spare resources to investigate what has changed. Perhaps another team member who has done work on the init system for antiX-19 might be able to get to it before me.
Thank you for all your time and effort.
I hope antiX developers will be able to get to resolving this issue with antiX 19, particularly as it affects across the board whoever is connecting SMB shares, not just me…Live antiX Boot Options (Previously posted by Xecure):
https://antixlinuxfan.miraheze.org/wiki/Table_of_antiX_Boot_Parameters
anticapitalista
@SamK – are you referring to the umountnfs-alternative fix?
The fix us still present on antiX-19 series including 19.2
Philosophers have interpreted the world in many ways; the point is to change it.
antiX with runit - leaner and meaner.
@SamK – are you referring to the umountnfs-alternative fix?
The fix us still present on antiX-19 series including 19.2
Yes, that’s the one.
During the reboot/shutdown process a message is displayed “stopping enhanced syslogd: rsyslogd”. After a timeout period the process completes normally. That is exactly the same as the symptom fixed by umountnfs-alternative.
In 2014 it was established the cause stems from the sequence in which the (then Debian) init.d system stopped services. The short version is, wireless networking was stopped before shares were unmounted. In turn that generated the timeout delay later in reboot/shutdown process.
Executing the link in either rc0.d or rc6.d to umountnfs-alternative earlier in the sequence prevented the pause condition and allowed the reboot/shutdown to continue without a delay.
The issue now occurs with both wired and wireless in antiX-19.2 but not in antiX-17.4.2.
PHerrmann
Hello!
If this may help :
I use Connectshares to mount Samba local network shares (NAS).
I encountered turnoff process halting after the line : “Stopping enhanced syslogd : rsyslogd” during 1 to 2 minutes, then resuming to full stop. This behavior was the same for 32 bit and 64 bit PC, both running antiX 19.2.
I found that, when disconnecting manually the shares using Disconnectshares before turnoff, the turnoff process was very quick, as it should be.
This behavior is reproducible.anticapitalista
The only difference that may be the cause is in /etc/init.d/.depend.stop
antiX-17.4.1 has this
TARGETS = umountnfs-alternative.sh ifplugd tlp virtualbox-guest-utils gpm haveged smartmontools slim alsa-utils resolvconf urandom sendsigs rsyslog umountnfs.sh rpcbind nfs-common hwclock.sh umountfs cryptdisks cryptdisks-early udev umountroot halt reboot sendsigs: alsa-utils gpm haveged ifplugd slim tlp virtualbox-guest-utils rsyslog: gpm haveged sendsigs smartmontools umountnfs.sh: alsa-utils gpm haveged ifplugd slim tlp virtualbox-guest-utils sendsigs rsyslog rpcbind: umountnfs-alternative.sh umountnfs.sh nfs-common: umountnfs-alternative.sh umountnfs.sh hwclock.sh: rsyslog nfs-common umountfs: alsa-utils gpm haveged ifplugd resolvconf slim tlp umountnfs-alternative.sh urandom virtualbox-guest-utils umountnfs.sh rpcbind hwclock.sh cryptdisks: umountfs cryptdisks-early: cryptdisks umountfs udev: cryptdisks cryptdisks-early umountroot: udev umountfs cryptdisks cryptdisks-early halt: umountroot reboot: umountrootantiX-19.2 has this
TARGETS = umountnfs-alternative.sh tlp virtualbox-guest-utils gpm haveged smartmontools connman elogind slim alsa-utils brightness resolvconf urandom ufw sendsigs rsyslog umountnfs.sh rpcbind nfs-common hwclock.sh umountfs cryptdisks cryptdisks-early udev umountroot halt reboot sendsigs: alsa-utils connman elogind gpm haveged slim tlp virtualbox-guest-utils rsyslog: sendsigs umountnfs.sh: rsyslog sendsigs alsa-utils connman elogind gpm haveged slim tlp virtualbox-guest-utils rpcbind: umountnfs.sh umountnfs-alternative.sh nfs-common: umountnfs.sh umountnfs-alternative.sh hwclock.sh: nfs-common rsyslog umountfs: hwclock.sh rpcbind umountnfs.sh alsa-utils brightness connman elogind gpm haveged resolvconf slim tlp umountnfs-alternative.sh urandom virtualbox-guest-utils cryptdisks: umountfs cryptdisks-early: cryptdisks umountfs udev: cryptdisks-early umountroot: udev cryptdisks-early cryptdisks umountfs halt: umountroot reboot: umountrootCould someone who uses connectshares on antiX-19 series, try changing to these two lines.
rpcbind: umountnfs-alternative.sh umountnfs.sh nfs-common: umountnfs-alternative.sh umountnfs.shPhilosophers have interpreted the world in many ways; the point is to change it.
antiX with runit - leaner and meaner.
umountnfs-alternative.sh is not the source of the problem.
Connman is the cause of the unwanted behaviour.
I do not have sufficient spare capacity provide more than superficial input.
Test Requirements
-
* No files in /etc/init.d modified by user
* WiFi connected via connman
* Remote shares can be mounted on local system
* Delay experienced during shutdown (rsyslogd)Test on Live System
-
* Boot with all persistence
* Uninstall Connman completely (i.e. purge all connman files & settings)
* Ensure no references exist to Conmann in the init.d system (ls /etc/init.d)
* Set up wifi using Ceni
* Mount remote shares on local system & read a few files to verify working
* Save root persistence (via Control Centre->Live)
* Power off local system while remote shares are still mountedExpected Outcome
-
* Local system shuts down without delay
Conclusion
By removing Connman and replacing it with Ceni, the conditions that are known to work in antiX-17 are recreated.When tested here, the above returns the normal, uninterupted shutdown behaviour. It demonstrates the manner in which Connman is stopped is the seat of the problem.
The init.d system provides a mechanism by which the killing of services can be sequenced. Connman is being killed too early in the shutdown routines. In turn that kills wifi before the remote and virtual filesystems can be unmounted, producing the delay to rsyslogd.
Fix
The fix is to ensure Connman is killed after umountnfs-alternative.sh.My testing, following SamK’s diagnosis, confirms antiX shuts down normally (does not hang on shutdown) when Ceni is used (Connman uninstalled).
Looks like the root of this issue has been identified. Thanks SamK!
So is there a parameter that can be specified in config files to prevent premature killing of Connman during shutdown?Live antiX Boot Options (Previously posted by Xecure):
https://antixlinuxfan.miraheze.org/wiki/Table_of_antiX_Boot_ParametersSo is there a parameter that can be specified in config files to prevent premature killing of Connman during shutdown?
No.
The fix requires a system level change and is best issued by the update/upgrade mechanism after testing and packaging. I do not have the spare resources to carry out the work, perhaps another team member might.
dolphin_oracle
tricky…
maybe adding Should-Stop: connman to the umountnfs-alternative.sh init script would be appropriate.
- This reply was modified 2 years, 11 months ago by dolphin_oracle.
maybe adding Should-Stop: connman to the umountnfs-alternative.sh init script would be appropriate.
Unfortunately not. connman and umountnfs-alternative.sh have the same kill priority i.e. the highest. Due to alphabetical order, connman is always killed before umountnfs-alternative.sh, so adding “Should-Stop: connman” is not a feasible approach.
dolphin_oracle
well, it worked ok for me.
test files attached. I added a echo to a /etc/umount.log into both init scripts, but only umountnfs-alternative.sh has the Should-Stop entry for connman.
nice to see you sam 🙂
- This reply was modified 2 years, 11 months ago by dolphin_oracle.
dolphin_oracle
checking out the /etc/init.d/.depend-stop file after running update-rc.d on the modified umountnfs-alternative.sh init file. note that connman has a dependent stop on umountnfs-alternative.sh
TARGETS = umountnfs-alternative.sh tlp unattended-upgrades gpm haveged irqbalance acpi-support smartmontools bluetooth dundee elogind ofono slim avahi-daemon saned cups alsa-utils brightness resolvconf urandom ufw connman sendsigs rsyslog umountnfs.sh rpcbind nfs-common hwclock.sh umountfs cryptdisks cryptdisks-early udev umountroot halt reboot avahi-daemon: saned connman: umountnfs-alternative.sh sendsigs: connman alsa-utils bluetooth dundee elogind gpm haveged irqbalance ofono saned slim tlp unattended-upgrades avahi-daemon rsyslog: sendsigs umountnfs.sh: rsyslog sendsigs connman alsa-utils bluetooth dundee elogind gpm haveged irqbalance ofono saned slim tlp unattended-upgrades avahi-daemon rpcbind: umountnfs.sh umountnfs-alternative.sh nfs-common: umountnfs.sh umountnfs-alternative.sh hwclock.sh: nfs-common rsyslog umountfs: hwclock.sh rpcbind umountnfs.sh connman alsa-utils bluetooth brightness dundee elogind gpm haveged irqbalance ofono resolvconf saned slim tlp umountnfs-alternative.sh unattended-upgrades urandom avahi-daemon cryptdisks: umountfs cryptdisks-early: cryptdisks umountfs udev: cryptdisks-early umountroot: udev cryptdisks-early cryptdisks umountfs halt: umountroot reboot: umountroot- This reply was modified 2 years, 11 months ago by dolphin_oracle.
nice to see you sam
Thanks d_o. I wish I could be around more often.
well, it worked ok for me.
OK, useful to know. I will have a look. There are a couple of things bothering me about this approach. I will be in touch to chat them over.
anticapitalista
I sent bugfix debs to the epos. Should appear soon.
desktop-defaults-core-antix has the new umountnfs-alternative.sh and connman a minor edit to its /etc/init.d/connman file (thanks d_o and SamK)
Philosophers have interpreted the world in many ways; the point is to change it.
antiX with runit - leaner and meaner.
- You must be logged in to reply to this topic.
