Re: case study - journalctl - where is logger output
On Wed, Sep 17, 2014 at 10:26 PM, Chris Murphy wrote: > On Sep 17, 2014, at 9:33 AM, Tom H wrote: >> >> AIUI, imjournal needs a file for rsyslog to read journald logs so it >> won't work with "Storage=none". >> >> To forward to rsyslog you have to: >> >> - use "Storage=volatile" (if you don't want to use auto/persistent) in >> "/etc/systemd/journald.conf" so journald will log to >> "/run/systemd/journal/" and imjournal will pick up logs from there >> >> or >> >> - use "$AddUnixListenSocket /run/systemd/journal/syslog" in >> "/etc/rsyslog.d/sd-socket.conf" so journald will log to >> "/run/systemd/journal/syslog" and imuxsocket will pick up logs from >> there (untested) > > My understanding is the current versions (Fedora 19+) of rsyslog and systemd > "just work" - rsyslog connects to systemd-journald.socket. I only skimmed > this thread but thought I saw something about a socket being disabled with > systemctl, and if it was the systemd-journald.socket well then that'd be a > reason why rsyslog isn't getting messages. By default, as long as "Storage=none" isn't set, the forward to rsyslog works. > But I'm not sure what all Balint has changed from the Fedora 20 defaults, > but my test VM is pristine so either Balint has changed something I haven't > changed and he hasn't changed it back to the default. Or maybe he's run into > a bug where he changed something, changed it back but it's not taking. Or > other… > > My test was strictly two things. The first was: > 1. dnf install rsyslog ### because it's not installed by default > 2. edited journald.conf changing Storage=auto to none > 3. Reboot > > There is no persistence at all in either /run/log/journal or /var/log/journal, > yet there are entries in both journalctl and /var/log/messages, and the logger > command inserts whatever I type after it into both. > > And the other test following this was: > > 4. change journald.conf Storage=auto again, i.e. back to default. > 5. delete /var/log/journal > 6. reboot > > And now I get a journal stored in /run/log/journal; from prior testing a while > ago this is the same behavior I get when Storage=volatile > > I didn't extensively test the difference between Storage=none and Storage=auto > or volatile (with /run/log/journal and /var/log/journal deleted). And I can't > really tell from the documentation or the resulting logs; but I'm going to > guess that none means it's using some limited amount of RAM (?) whereas =auto > or volatile using /run/log/journal is on tmpfs and after it uses up the > 50% RAM limit it starts to swap to disk. In either case they're > toast at the next reboot. I've just done the same two tests with the same results. From "man journald.conf: "none" turns off all storage, all log data received will be dropped. Forwarding to other targets, such as the console, the kernel log buffer or a syslog daemon will still work however. What it doesn't say is whether "imjournal" will work or not. In both our tests, imjournal (the default rsyslog module for journald) doesn't work when "Storage=none" is set so you need to use imuxsocket. I've also tested "$AddUnixListenSocket /run/systemd/journal/syslog" and logs are forwarded to rsyslog. -- users mailing list users@lists.fedoraproject.org To unsubscribe or change subscription options: https://admin.fedoraproject.org/mailman/listinfo/users Fedora Code of Conduct: http://fedoraproject.org/code-of-conduct Guidelines: http://fedoraproject.org/wiki/Mailing_list_guidelines Have a question? Ask away: http://ask.fedoraproject.org
Re: case study - journalctl - where is logger output
On Sep 17, 2014, at 9:33 AM, Tom H wrote: > On Wed, Sep 17, 2014 at 6:40 AM, Balint Szigeti wrote: >> >> Well, I have definitely a problem: >> >> [root@szigeti-6560b ~]# vi /etc/systemd/journald.conf >> [root@szigeti-6560b ~]# grep none $_ >> Storage=none >> [root@szigeti-6560b ~]# systemctl restart systemd-journald.service >> [root@szigeti-6560b ~]# logger foo >> [root@szigeti-6560b ~]# grep foo /var/log/messages >> [root@szigeti-6560b ~]# journalctl -n 100 |grep foo >> [root@szigeti-6560b ~]# >> [root@szigeti-6560b ~]# vi /etc/systemd/journald.conf >> [root@szigeti-6560b ~]# systemctl restart systemd-journald.service >> [root@szigeti-6560b ~]# grep Storage /etc/systemd/journald.conf >> #Storage=none >> #Storage=auto >> [root@szigeti-6560b ~]# logger AAA >> [root@szigeti-6560b ~]# grep AA /var/log/messages >> Sep 17 11:38:14 SZIGETI-6560B bszigeti: AAA >> [root@szigeti-6560b ~]# journalctl -n 100 |grep >> Sep 17 11:38:14 szigeti-6560b bszigeti[28839]: AAA > > AIUI, imjournal needs a file for rsyslog to read journald logs so it > won't work with "Storage=none". > > To forward to rsyslog you have to: > > - use "Storage=volatile" (if you don't want to use auto/persistent) in > "/etc/systemd/journald.conf" so journald will log to > "/run/systemd/journal/" and imjournal will pick up logs from there > > or > > - use "$AddUnixListenSocket /run/systemd/journal/syslog" in > "/etc/rsyslog.d/sd-socket.conf" so journald will log to > "/run/systemd/journal/syslog" and imuxsocket will pick up logs from > there (untested) Take this with a grain of salt, I confuse easily… My understanding is the current versions (Fedora 19+) of rsyslog and systemd "just work" - rsyslog connects to systemd-journald.socket. I only skimmed this thread but thought I saw something about a socket being disabled with systemctl, and if it was the systemd-journald.socket well then that'd be a reason why rsyslog isn't getting messages. But I'm not sure what all Balint has changed from the Fedora 20 defaults, but my test VM is pristine so either Balint has changed something I haven't changed and he hasn't changed it back to the default. Or maybe he's run into a bug where he changed something, changed it back but it's not taking. Or other… My test was strictly two things. The first was: 1. dnf install rsyslog ###because it's not installed by default 2. edited journald.conf changing Storage=auto to none 3. Reboot There is no persistence at all in either /run/log/journal or /var/log/journal, yet there are entries in both journalctl and /var/log/messages, and the logger command inserts whatever I type after it into both. And the other test following this was: 4. change journald.conf Storage=auto again, i.e. back to default. 5. delete /var/log/journal 6. reboot And now I get a journal stored in /run/log/journal; from prior testing a while ago this is the same behavior I get when Storage=volatile I didn't extensively test the difference between Storage=none and Storage=auto or volatile (with /run/log/journal and /var/log/journal deleted). And I can't really tell from the documentation or the resulting logs; but I'm going to guess that none means it's using some limited amount of RAM (?) whereas =auto or volatile using /run/log/journal is on tmpfs and after it uses up the 50% RAM limit it starts to swap to disk. In either case they're toast at the next reboot. Chris Murphy -- users mailing list users@lists.fedoraproject.org To unsubscribe or change subscription options: https://admin.fedoraproject.org/mailman/listinfo/users Fedora Code of Conduct: http://fedoraproject.org/code-of-conduct Guidelines: http://fedoraproject.org/wiki/Mailing_list_guidelines Have a question? Ask away: http://ask.fedoraproject.org
Re: case study - journalctl - where is logger output
On Wed, Sep 17, 2014 at 6:40 AM, Balint Szigeti wrote: > > Well, I have definitely a problem: > > [root@szigeti-6560b ~]# vi /etc/systemd/journald.conf > [root@szigeti-6560b ~]# grep none $_ > Storage=none > [root@szigeti-6560b ~]# systemctl restart systemd-journald.service > [root@szigeti-6560b ~]# logger foo > [root@szigeti-6560b ~]# grep foo /var/log/messages > [root@szigeti-6560b ~]# journalctl -n 100 |grep foo > [root@szigeti-6560b ~]# > [root@szigeti-6560b ~]# vi /etc/systemd/journald.conf > [root@szigeti-6560b ~]# systemctl restart systemd-journald.service > [root@szigeti-6560b ~]# grep Storage /etc/systemd/journald.conf > #Storage=none > #Storage=auto > [root@szigeti-6560b ~]# logger AAA > [root@szigeti-6560b ~]# grep AA /var/log/messages > Sep 17 11:38:14 SZIGETI-6560B bszigeti: AAA > [root@szigeti-6560b ~]# journalctl -n 100 |grep > Sep 17 11:38:14 szigeti-6560b bszigeti[28839]: AAA AIUI, imjournal needs a file for rsyslog to read journald logs so it won't work with "Storage=none". To forward to rsyslog you have to: - use "Storage=volatile" (if you don't want to use auto/persistent) in "/etc/systemd/journald.conf" so journald will log to "/run/systemd/journal/" and imjournal will pick up logs from there or - use "$AddUnixListenSocket /run/systemd/journal/syslog" in "/etc/rsyslog.d/sd-socket.conf" so journald will log to "/run/systemd/journal/syslog" and imuxsocket will pick up logs from there (untested) -- users mailing list users@lists.fedoraproject.org To unsubscribe or change subscription options: https://admin.fedoraproject.org/mailman/listinfo/users Fedora Code of Conduct: http://fedoraproject.org/code-of-conduct Guidelines: http://fedoraproject.org/wiki/Mailing_list_guidelines Have a question? Ask away: http://ask.fedoraproject.org
Re: case study - journalctl - where is logger output
On Tue, 2014-09-16 at 17:46 -0600, Chris Murphy wrote: > On Sep 16, 2014, at 9:11 AM, Balint Szigeti wrote: > > > On Tue, 2014-09-16 at 09:52 -0400, Tom H wrote: > >> On Sat, Sep 13, 2014 at 11:47 AM, Balint Szigeti >> > wrote: > >> > >> > On Thu, 2014-09-11 at 16:16 -0400, Tom H wrote: > >> > > >> > On Thu, Sep 11, 2014 at 2:04 PM, Balint Szigeti > >> > wrote: > >> > > >> > > >> >> today I installed the rsyslog and enable it then disabled (then masked) > >> >> systemd-journal-flush, systemd-journald services. Plus I disabled > >> >> systemd-journald.socket as well. > >> >> It broke my system. After I closed the sudo session I could gain root > >> >> access > >> >> plus I couldn't start any program only forks for the existed ones (like > >> >> gnome terminal). > >> >> The reboot didn't work. The box just didn't start up. :( (just remark - > >> >> systemd is not depends on itself) > >> > > >> > I disabled all of the journal service and socket units and rebooted > >> > without a hitch. It was in an X-less VM though so perhaps things go > >> > awry when booting a DE (I don't see why it whould). > >> > > >> > > >> >> I booted into runlevel 1 (yeeeah - runlevel doesn't exist on systemd - I > >> >> wanted to say rescue.target) and redo the mask and enable everything. > >> > > >> > I boot into runlevel 1 when I use "1" on the kernel cmdline. > >> > > >> > > >> >> I've noticed the rsyslog doesn't listen to the system logging. > >> >> > >> >> I've run logger command but I don't find it in the log. I've checked the > >> >> journalctl and /var/log/messages file as well. > >> >> > >> >> # logger -t hello > >> >> # journalctl |grep hello > >> >> # grep hello /var/log/messages > >> >> # > >> > > >> > Same here. > >> > > >> > Is journald supposed to be turned off when using systemd? Why do you > >> > want it off? You can set "Storage=volatile" in > >> > "/etc/systemd/journald.conf" and 1) you'll only have rsyslog logs > >> > across reboots and 2) the journald logs will be written to the > >> > "/run/log/journal/" tmpfs so journald will simply collect logs for > >> > rsyslog. > >> > > >> > I've just tried to set the Storage entry in /etc/systemd/journald.conf > >> > to > >> > "none" according to manual page and off course no effect. > >> > Also tried to set LogTarget to "syslog" in /etc/systemd/system.conf and > >> > reboot (funny, hurray we become Windows..)but no effect. > >> > > >> > Needless to say, the logs are being found in journalctl and messages > >> > file of > >> > course. I don't think to raise bug because any time when they hear > >> > someone > >> > doesn't want to use their 'solutions' they refuse/ignore or set the > >> > ticket > >> > to WONTFIX. > >> > Personally that is my bigger problem. > >> > > >> > I finished testing the syslog. I think the only thing that we can do to > >> > accept and shut our mouth :( > >> > >> > >> It's working here. systemd must sense that you don't like it and it's > >> messin' wit' u! > >> > >> > >> > > you are right. I don't like it. > >> # grep Storage /etc/systemd/journald.conf > >> Storage=none > >> > > I've set it and restarted the "systemctl restart systemd-journald.service" > > journald and the system stopped to logging. :( > > > From defaults, only changing journald.conf Storage=none, installing and > enabling rsyslog, I can't reproduce the problem on either Fedora 20 or 21. > > [root@f20v log]# logger biscuit > [root@f20v log]# grep biscuit /var/log/messages > Sep 16 10:33:42 f20v root: biscuit > [root@f20v log]# journalctl | grep biscuit > Sep 16 10:33:42 f20v.localdomain root[786]: biscuit > > rsyslog-7.4.8-1.fc20.1.x86_64 > systemd-208-21.fc20.x86_64 Well, I have definitely a problem: [root@szigeti-6560b ~]# vi /etc/systemd/journald.conf [root@szigeti-6560b ~]# grep none $_ Storage=none [root@szigeti-6560b ~]# systemctl restart systemd-journald.service [root@szigeti-6560b ~]# logger foo [root@szigeti-6560b ~]# grep foo /var/log/messages [root@szigeti-6560b ~]# journalctl -n 100 |grep foo [root@szigeti-6560b ~]# [root@szigeti-6560b ~]# vi /etc/systemd/journald.conf [root@szigeti-6560b ~]# systemctl restart systemd-journald.service [root@szigeti-6560b ~]# grep Storage /etc/systemd/journald.conf #Storage=none #Storage=auto [root@szigeti-6560b ~]# logger AAA [root@szigeti-6560b ~]# grep AA /var/log/messages Sep 17 11:38:14 SZIGETI-6560B bszigeti: AAA [root@szigeti-6560b ~]# journalctl -n 100 |grep Sep 17 11:38:14 szigeti-6560b bszigeti[28839]: AAA rsyslog-7.4.8-1.fc20.1.x86_64 systemd-208-21.fc20.x86_64 Do you have any clue? > > > > Chris Murphy > -- users mailing list users@lists.fedoraproject.org To unsubscribe or change subscription options: https://admin.fedoraproject.org/mailman/listinfo/users Fedora Code of Conduct: http://fedoraproject.org/code-of-conduct Guidelines: http://fedoraproject.org/wiki/Mailing_list_guidelines Have a question? Ask away: http://ask.fedoraproject.o
Re: case study - journalctl - where is logger output
On Sep 16, 2014, at 9:11 AM, Balint Szigeti wrote: > On Tue, 2014-09-16 at 09:52 -0400, Tom H wrote: >> On Sat, Sep 13, 2014 at 11:47 AM, Balint Szigeti > > wrote: >> >> > On Thu, 2014-09-11 at 16:16 -0400, Tom H wrote: >> > >> > On Thu, Sep 11, 2014 at 2:04 PM, Balint Szigeti >> > wrote: >> > >> > >> >> today I installed the rsyslog and enable it then disabled (then masked) >> >> systemd-journal-flush, systemd-journald services. Plus I disabled >> >> systemd-journald.socket as well. >> >> It broke my system. After I closed the sudo session I could gain root >> >> access >> >> plus I couldn't start any program only forks for the existed ones (like >> >> gnome terminal). >> >> The reboot didn't work. The box just didn't start up. :( (just remark - >> >> systemd is not depends on itself) >> > >> > I disabled all of the journal service and socket units and rebooted >> > without a hitch. It was in an X-less VM though so perhaps things go >> > awry when booting a DE (I don't see why it whould). >> > >> > >> >> I booted into runlevel 1 (yeeeah - runlevel doesn't exist on systemd - I >> >> wanted to say rescue.target) and redo the mask and enable everything. >> > >> > I boot into runlevel 1 when I use "1" on the kernel cmdline. >> > >> > >> >> I've noticed the rsyslog doesn't listen to the system logging. >> >> >> >> I've run logger command but I don't find it in the log. I've checked the >> >> journalctl and /var/log/messages file as well. >> >> >> >> # logger -t hello >> >> # journalctl |grep hello >> >> # grep hello /var/log/messages >> >> # >> > >> > Same here. >> > >> > Is journald supposed to be turned off when using systemd? Why do you >> > want it off? You can set "Storage=volatile" in >> > "/etc/systemd/journald.conf" and 1) you'll only have rsyslog logs >> > across reboots and 2) the journald logs will be written to the >> > "/run/log/journal/" tmpfs so journald will simply collect logs for >> > rsyslog. >> > >> > I've just tried to set the Storage entry in /etc/systemd/journald.conf to >> > "none" according to manual page and off course no effect. >> > Also tried to set LogTarget to "syslog" in /etc/systemd/system.conf and >> > reboot (funny, hurray we become Windows..)but no effect. >> > >> > Needless to say, the logs are being found in journalctl and messages file >> > of >> > course. I don't think to raise bug because any time when they hear someone >> > doesn't want to use their 'solutions' they refuse/ignore or set the ticket >> > to WONTFIX. >> > Personally that is my bigger problem. >> > >> > I finished testing the syslog. I think the only thing that we can do to >> > accept and shut our mouth :( >> >> >> It's working here. systemd must sense that you don't like it and it's >> messin' wit' u! >> >> >> > you are right. I don't like it. >> # grep Storage /etc/systemd/journald.conf >> Storage=none >> > I've set it and restarted the "systemctl restart systemd-journald.service" > journald and the system stopped to logging. :( From defaults, only changing journald.conf Storage=none, installing and enabling rsyslog, I can't reproduce the problem on either Fedora 20 or 21. [root@f20v log]# logger biscuit [root@f20v log]# grep biscuit /var/log/messages Sep 16 10:33:42 f20v root: biscuit [root@f20v log]# journalctl | grep biscuit Sep 16 10:33:42 f20v.localdomain root[786]: biscuit rsyslog-7.4.8-1.fc20.1.x86_64 systemd-208-21.fc20.x86_64 Chris Murphy -- users mailing list users@lists.fedoraproject.org To unsubscribe or change subscription options: https://admin.fedoraproject.org/mailman/listinfo/users Fedora Code of Conduct: http://fedoraproject.org/code-of-conduct Guidelines: http://fedoraproject.org/wiki/Mailing_list_guidelines Have a question? Ask away: http://ask.fedoraproject.org
Re: case study - journalctl - where is logger output
On Tue, Sep 16, 2014 at 11:11 AM, Balint Szigeti wrote: > > I don't have jc command. It's just an alias for "journalctl". -- users mailing list users@lists.fedoraproject.org To unsubscribe or change subscription options: https://admin.fedoraproject.org/mailman/listinfo/users Fedora Code of Conduct: http://fedoraproject.org/code-of-conduct Guidelines: http://fedoraproject.org/wiki/Mailing_list_guidelines Have a question? Ask away: http://ask.fedoraproject.org
Re: case study - journalctl - where is logger output
On Tue, 2014-09-16 at 09:52 -0400, Tom H wrote: > On Sat, Sep 13, 2014 at 11:47 AM, Balint Szigeti > wrote: > > On Thu, 2014-09-11 at 16:16 -0400, Tom H wrote: > > > > On Thu, Sep 11, 2014 at 2:04 PM, Balint Szigeti > > wrote: > > > > > >> today I installed the rsyslog and enable it then disabled (then masked) > >> systemd-journal-flush, systemd-journald services. Plus I disabled > >> systemd-journald.socket as well. > >> It broke my system. After I closed the sudo session I could gain root > >> access > >> plus I couldn't start any program only forks for the existed ones (like > >> gnome terminal). > >> The reboot didn't work. The box just didn't start up. :( (just remark - > >> systemd is not depends on itself) > > > > I disabled all of the journal service and socket units and rebooted > > without a hitch. It was in an X-less VM though so perhaps things go > > awry when booting a DE (I don't see why it whould). > > > > > >> I booted into runlevel 1 (yeeeah - runlevel doesn't exist on systemd - I > >> wanted to say rescue.target) and redo the mask and enable everything. > > > > I boot into runlevel 1 when I use "1" on the kernel cmdline. > > > > > >> I've noticed the rsyslog doesn't listen to the system logging. > >> > >> I've run logger command but I don't find it in the log. I've checked the > >> journalctl and /var/log/messages file as well. > >> > >> # logger -t hello > >> # journalctl |grep hello > >> # grep hello /var/log/messages > >> # > > > > Same here. > > > > Is journald supposed to be turned off when using systemd? Why do you > > want it off? You can set "Storage=volatile" in > > "/etc/systemd/journald.conf" and 1) you'll only have rsyslog logs > > across reboots and 2) the journald logs will be written to the > > "/run/log/journal/" tmpfs so journald will simply collect logs for > > rsyslog. > > > > I've just tried to set the Storage entry in /etc/systemd/journald.conf to > > "none" according to manual page and off course no effect. > > Also tried to set LogTarget to "syslog" in /etc/systemd/system.conf and > > reboot (funny, hurray we become Windows..)but no effect. > > > > Needless to say, the logs are being found in journalctl and messages file of > > course. I don't think to raise bug because any time when they hear someone > > doesn't want to use their 'solutions' they refuse/ignore or set the ticket > > to WONTFIX. > > Personally that is my bigger problem. > > > > I finished testing the syslog. I think the only thing that we can do to > > accept and shut our mouth :( > > It's working here. systemd must sense that you don't like it and it's > messin' wit' u! you are right. I don't like it. > > # grep Storage /etc/systemd/journald.conf > Storage=none I've set it and restarted the "systemctl restart systemd-journald.service" journald and the system stopped to logging. :( # systemctl restart systemd-journald.service # logger helo # journalctl -n 3 -- Logs begin at Thu 2014-09-11 18:34:22 BST, end at Tue 2014-09-16 16:08:03 BST. -- Sep 16 16:01:01 szigeti-6560b run-parts[30639]: (/etc/cron.hourly) finished mcelog.cron Sep 16 16:08:03 szigeti-6560b systemd[1]: Stopping Journal Service... Sep 16 16:08:03 szigeti-6560b systemd-journal[705]: Journal stopped # grep helo /var/log/messages # > > # ll /var/log/journal/ > total 0 > > # ll /run/systemd/journal/ > total 4 > srw-rw-rw-. 1 root root 0 Sep 16 09:45 dev-log > -rw-r--r--. 1 root root 0 Sep 16 09:45 flushed > -rw-r--r--. 1 root root 8 Sep 16 09:45 kernel-seqnum > srw-rw-rw-. 1 root root 0 Sep 16 09:45 socket > srw-rw-rw-. 1 root root 0 Sep 16 09:45 stdout > I don't have jc command. > # jc --list-boots > 0 db26ae69f1714eafb3058bcbc8d257c6 Tue 2014-09-16 09:45:09 EDT—Tue > 2014-09-16 09:45:11 EDT > > # jc > -- Logs begin at Tue 2014-09-16 09:45:09 EDT, end at Tue 2014-09-16 > 09:45:11 EDT. -- > Sep 16 09:45:09 localhost systemd-journal[86]: Runtime journal is > using 6.0M (max allowed 48.6M, trying to leave 73.0M free of 480.7M > available → current limit 48.6M). > Sep 16 09:45:09 localhost systemd-journal[86]: Runtime journal is > using 6.0M (max allowed 48.6M, trying to leave 73.0M free of 480.7M > available → current limit 48.6M). > Sep 16 09:45:09 localhost kernel: Initializing cgroup subsys cpuset > Sep 16 09:45:09 localhost kernel: Initializing cgroup subsys cpu > Sep 16 09:45:09 localhost kernel: Initializing cgroup subsys cpuacct > Sep 16 09:45:09 localhost kernel: Linux version > 3.17.0-0.rc4.git4.1.fc22.x86_64 > (mockbu...@bkernel01.phx2.fedoraproject.org) (gcc version 4.9.1 > 20140813 (Red Hat 4.9.1-7) (GCC) ) #1 SMP Fri Sep 12 17:34:49 UTC 2014 > Sep 16 09:45:09 localhost kernel: Command line: > BOOT_IMAGE=/boot/vmlinuz-3.17.0-0.rc4.git4.1.fc22.x86_64 > root=UUID=5bac0a63-0755-4127-9521-ae59faee6327 ro > ... -- users mailing list users@lists.fedoraproject.org To unsubscribe or change subscription options: https://admin.fedoraproject.org/mailman/listinfo/users Fedora Code of Conduct: h
Re: case study - journalctl - where is logger output
On Sat, Sep 13, 2014 at 11:47 AM, Balint Szigeti wrote: > On Thu, 2014-09-11 at 16:16 -0400, Tom H wrote: > > On Thu, Sep 11, 2014 at 2:04 PM, Balint Szigeti > wrote: > > >> today I installed the rsyslog and enable it then disabled (then masked) >> systemd-journal-flush, systemd-journald services. Plus I disabled >> systemd-journald.socket as well. >> It broke my system. After I closed the sudo session I could gain root >> access >> plus I couldn't start any program only forks for the existed ones (like >> gnome terminal). >> The reboot didn't work. The box just didn't start up. :( (just remark - >> systemd is not depends on itself) > > I disabled all of the journal service and socket units and rebooted > without a hitch. It was in an X-less VM though so perhaps things go > awry when booting a DE (I don't see why it whould). > > >> I booted into runlevel 1 (yeeeah - runlevel doesn't exist on systemd - I >> wanted to say rescue.target) and redo the mask and enable everything. > > I boot into runlevel 1 when I use "1" on the kernel cmdline. > > >> I've noticed the rsyslog doesn't listen to the system logging. >> >> I've run logger command but I don't find it in the log. I've checked the >> journalctl and /var/log/messages file as well. >> >> # logger -t hello >> # journalctl |grep hello >> # grep hello /var/log/messages >> # > > Same here. > > Is journald supposed to be turned off when using systemd? Why do you > want it off? You can set "Storage=volatile" in > "/etc/systemd/journald.conf" and 1) you'll only have rsyslog logs > across reboots and 2) the journald logs will be written to the > "/run/log/journal/" tmpfs so journald will simply collect logs for > rsyslog. > > I've just tried to set the Storage entry in /etc/systemd/journald.conf to > "none" according to manual page and off course no effect. > Also tried to set LogTarget to "syslog" in /etc/systemd/system.conf and > reboot (funny, hurray we become Windows..)but no effect. > > Needless to say, the logs are being found in journalctl and messages file of > course. I don't think to raise bug because any time when they hear someone > doesn't want to use their 'solutions' they refuse/ignore or set the ticket > to WONTFIX. > Personally that is my bigger problem. > > I finished testing the syslog. I think the only thing that we can do to > accept and shut our mouth :( It's working here. systemd must sense that you don't like it and it's messin' wit' u! # grep Storage /etc/systemd/journald.conf Storage=none # ll /var/log/journal/ total 0 # ll /run/systemd/journal/ total 4 srw-rw-rw-. 1 root root 0 Sep 16 09:45 dev-log -rw-r--r--. 1 root root 0 Sep 16 09:45 flushed -rw-r--r--. 1 root root 8 Sep 16 09:45 kernel-seqnum srw-rw-rw-. 1 root root 0 Sep 16 09:45 socket srw-rw-rw-. 1 root root 0 Sep 16 09:45 stdout # jc --list-boots 0 db26ae69f1714eafb3058bcbc8d257c6 Tue 2014-09-16 09:45:09 EDT—Tue 2014-09-16 09:45:11 EDT # jc -- Logs begin at Tue 2014-09-16 09:45:09 EDT, end at Tue 2014-09-16 09:45:11 EDT. -- Sep 16 09:45:09 localhost systemd-journal[86]: Runtime journal is using 6.0M (max allowed 48.6M, trying to leave 73.0M free of 480.7M available → current limit 48.6M). Sep 16 09:45:09 localhost systemd-journal[86]: Runtime journal is using 6.0M (max allowed 48.6M, trying to leave 73.0M free of 480.7M available → current limit 48.6M). Sep 16 09:45:09 localhost kernel: Initializing cgroup subsys cpuset Sep 16 09:45:09 localhost kernel: Initializing cgroup subsys cpu Sep 16 09:45:09 localhost kernel: Initializing cgroup subsys cpuacct Sep 16 09:45:09 localhost kernel: Linux version 3.17.0-0.rc4.git4.1.fc22.x86_64 (mockbu...@bkernel01.phx2.fedoraproject.org) (gcc version 4.9.1 20140813 (Red Hat 4.9.1-7) (GCC) ) #1 SMP Fri Sep 12 17:34:49 UTC 2014 Sep 16 09:45:09 localhost kernel: Command line: BOOT_IMAGE=/boot/vmlinuz-3.17.0-0.rc4.git4.1.fc22.x86_64 root=UUID=5bac0a63-0755-4127-9521-ae59faee6327 ro ... -- users mailing list users@lists.fedoraproject.org To unsubscribe or change subscription options: https://admin.fedoraproject.org/mailman/listinfo/users Fedora Code of Conduct: http://fedoraproject.org/code-of-conduct Guidelines: http://fedoraproject.org/wiki/Mailing_list_guidelines Have a question? Ask away: http://ask.fedoraproject.org
Re: case study - journalctl - where is logger output
On Sep 13, 2014, at 9:47 AM, Balint Szigeti wrote: > > I've just tried to set the Storage entry in /etc/systemd/journald.conf to > "none" according to manual page and off course no effect. > Also tried to set LogTarget to "syslog" in /etc/systemd/system.conf and > reboot (funny, hurray we become Windows..)but no effect. > > Needless to say, the logs are being found in journalctl and messages file of > course. I don't think to raise bug because any time when they hear someone > doesn't want to use their 'solutions' they refuse/ignore or set the ticket to > WONTFIX. I don't understand the problem. You're saying logs are found in journalctl and messages, so what's the bug/problem? systemd-journald is expected to always run because it knows a ton of what's going from the kernel and systemd itself. The traditional syslog needs to be systemd-journald compatible, and I'm pretty sure rsyslog is. It directly reads the stream from systemd-journald. http://www.freedesktop.org/wiki/Software/systemd/syslog/ http://blog.delouw.ch/2013/07/24/why-journalctl-is-cool-and-syslog-will-survive-for-another-decade/ If you merely delete /var/log/journal, then systemd-journald will log to /run/log/journal which is not persistent. And a journald compatible syslogger will read that stream, and log it to /var/log/messages just like before except it probably contains more information since it should contain all journald logged events before syslogd started. I don't think none is what you want because then syslog has nothing available to it to log. At least that's my understanding of it. Chris Murphy-- users mailing list users@lists.fedoraproject.org To unsubscribe or change subscription options: https://admin.fedoraproject.org/mailman/listinfo/users Fedora Code of Conduct: http://fedoraproject.org/code-of-conduct Guidelines: http://fedoraproject.org/wiki/Mailing_list_guidelines Have a question? Ask away: http://ask.fedoraproject.org
Re: case study - journalctl - where is logger output
On Thu, 2014-09-11 at 16:16 -0400, Tom H wrote: > On Thu, Sep 11, 2014 at 2:04 PM, Balint Szigeti wrote: > > > > today I installed the rsyslog and enable it then disabled (then masked) > > systemd-journal-flush, systemd-journald services. Plus I disabled > > systemd-journald.socket as well. > > It broke my system. After I closed the sudo session I could gain root access > > plus I couldn't start any program only forks for the existed ones (like > > gnome terminal). > > The reboot didn't work. The box just didn't start up. :( (just remark - > > systemd is not depends on itself) > > I disabled all of the journal service and socket units and rebooted > without a hitch. It was in an X-less VM though so perhaps things go > awry when booting a DE (I don't see why it whould). > > > > I booted into runlevel 1 (yeeeah - runlevel doesn't exist on systemd - I > > wanted to say rescue.target) and redo the mask and enable everything. > > I boot into runlevel 1 when I use "1" on the kernel cmdline. > > > > I've noticed the rsyslog doesn't listen to the system logging. > > > > I've run logger command but I don't find it in the log. I've checked the > > journalctl and /var/log/messages file as well. > > > > # logger -t hello > > # journalctl |grep hello > > # grep hello /var/log/messages > > # > > Same here. > > Is journald supposed to be turned off when using systemd? Why do you > want it off? You can set "Storage=volatile" in > "/etc/systemd/journald.conf" and 1) you'll only have rsyslog logs > across reboots and 2) the journald logs will be written to the > "/run/log/journal/" tmpfs so journald will simply collect logs for > rsyslog. I've just tried to set the Storage entry in /etc/systemd/journald.conf to "none" according to manual page and off course no effect. Also tried to set LogTarget to "syslog" in /etc/systemd/system.conf and reboot (funny, hurray we become Windows..)but no effect. Needless to say, the logs are being found in journalctl and messages file of course. I don't think to raise bug because any time when they hear someone doesn't want to use their 'solutions' they refuse/ignore or set the ticket to WONTFIX. Personally that is my bigger problem. I finished testing the syslog. I think the only thing that we can do to accept and shut our mouth :( Balint -- users mailing list users@lists.fedoraproject.org To unsubscribe or change subscription options: https://admin.fedoraproject.org/mailman/listinfo/users Fedora Code of Conduct: http://fedoraproject.org/code-of-conduct Guidelines: http://fedoraproject.org/wiki/Mailing_list_guidelines Have a question? Ask away: http://ask.fedoraproject.org
Re: case study - journalctl - where is logger output
On Thu, 2014-09-11 at 16:16 -0400, Tom H wrote: > On Thu, Sep 11, 2014 at 2:04 PM, Balint Szigeti wrote: > > > > today I installed the rsyslog and enable it then disabled (then masked) > > systemd-journal-flush, systemd-journald services. Plus I disabled > > systemd-journald.socket as well. > > It broke my system. After I closed the sudo session I could gain root access > > plus I couldn't start any program only forks for the existed ones (like > > gnome terminal). > > The reboot didn't work. The box just didn't start up. :( (just remark - > > systemd is not depends on itself) > > I disabled all of the journal service and socket units and rebooted > without a hitch. It was in an X-less VM though so perhaps things go > awry when booting a DE (I don't see why it whould). > > > > I booted into runlevel 1 (yeeeah - runlevel doesn't exist on systemd - I > > wanted to say rescue.target) and redo the mask and enable everything. > > I boot into runlevel 1 when I use "1" on the kernel cmdline. > > > > I've noticed the rsyslog doesn't listen to the system logging. > > > > I've run logger command but I don't find it in the log. I've checked the > > journalctl and /var/log/messages file as well. > > > > # logger -t hello > > # journalctl |grep hello > > # grep hello /var/log/messages > > # > > Same here. > > Is journald supposed to be turned off when using systemd? Why do you > want it off? You can set "Storage=volatile" in > "/etc/systemd/journald.conf" and 1) you'll only have rsyslog logs > across reboots and 2) the journald logs will be written to the > "/run/log/journal/" tmpfs so journald will simply collect logs for > rsyslog. I didn't know that. Thank you very much. Balint -- users mailing list users@lists.fedoraproject.org To unsubscribe or change subscription options: https://admin.fedoraproject.org/mailman/listinfo/users Fedora Code of Conduct: http://fedoraproject.org/code-of-conduct Guidelines: http://fedoraproject.org/wiki/Mailing_list_guidelines Have a question? Ask away: http://ask.fedoraproject.org
Re: case study - journalctl - where is logger output
On Thu, Sep 11, 2014 at 2:04 PM, Balint Szigeti wrote: > today I installed the rsyslog and enable it then disabled (then masked) > systemd-journal-flush, systemd-journald services. Plus I disabled > systemd-journald.socket as well. > It broke my system. After I closed the sudo session I could gain root access > plus I couldn't start any program only forks for the existed ones (like > gnome terminal). > The reboot didn't work. The box just didn't start up. :( (just remark - > systemd is not depends on itself) I disabled all of the journal service and socket units and rebooted without a hitch. It was in an X-less VM though so perhaps things go awry when booting a DE (I don't see why it whould). > I booted into runlevel 1 (yeeeah - runlevel doesn't exist on systemd - I > wanted to say rescue.target) and redo the mask and enable everything. I boot into runlevel 1 when I use "1" on the kernel cmdline. > I've noticed the rsyslog doesn't listen to the system logging. > > I've run logger command but I don't find it in the log. I've checked the > journalctl and /var/log/messages file as well. > > # logger -t hello > # journalctl |grep hello > # grep hello /var/log/messages > # Same here. Is journald supposed to be turned off when using systemd? Why do you want it off? You can set "Storage=volatile" in "/etc/systemd/journald.conf" and 1) you'll only have rsyslog logs across reboots and 2) the journald logs will be written to the "/run/log/journal/" tmpfs so journald will simply collect logs for rsyslog. -- users mailing list users@lists.fedoraproject.org To unsubscribe or change subscription options: https://admin.fedoraproject.org/mailman/listinfo/users Fedora Code of Conduct: http://fedoraproject.org/code-of-conduct Guidelines: http://fedoraproject.org/wiki/Mailing_list_guidelines Have a question? Ask away: http://ask.fedoraproject.org
Re: case study - journalctl - where is logger output
okay, I figured it out. If the journalctl doesn't run neither rsyslog nor journalctl (last one is obvious) don't get anything. after I started the journalctl the logger output appeared in journalclt and tail /var/log/messages output. Balint On Thu, 2014-09-11 at 19:04 +0100, Balint Szigeti wrote: > hello > > today I installed the rsyslog and enable it then disabled (then > masked) systemd-journal-flush, systemd-journald services. Plus I > disabled systemd-journald.socket as well. > It broke my system. After I closed the sudo session I could gain root > access plus I couldn't start any program only forks for the existed > ones (like gnome terminal). > The reboot didn't work. The box just didn't start up. :( (just remark > - systemd is not depends on itself) > > I booted into runlevel 1 (yeeeah - runlevel doesn't exist on systemd - > I wanted to say rescue.target) and redo the mask and enable > everything. > > I've noticed the rsyslog doesn't listen to the system logging. > > I've run logger command but I don't find it in the log. I've checked > the journalctl and /var/log/messages file as well. > > # logger -t hello > # journalctl |grep hello > # grep hello /var/log/messages > # > > Does anyone have a clue? -- users mailing list users@lists.fedoraproject.org To unsubscribe or change subscription options: https://admin.fedoraproject.org/mailman/listinfo/users Fedora Code of Conduct: http://fedoraproject.org/code-of-conduct Guidelines: http://fedoraproject.org/wiki/Mailing_list_guidelines Have a question? Ask away: http://ask.fedoraproject.org
