Announcement

Collapse
No announcement yet.

Fedora 20 Might No Longer Install Syslog

Collapse
X
  • Filter
  • Time
  • Show
Clear All
new posts

  • #16
    Originally posted by bkor View Post
    Then be clear already or shut up.
    Its dog slow? i mean it takes 30s for journald to just page to end and i have only 1049322 lines there
    /var/log/journal

    [root@localhost journal]# du -h
    502M ./fbb0c4896a9c0f680475c0a07edd2c5a

    journal data file is sure BIG no wonder its so slow does it contain copies of my encrypted files for "backup" purposes ? or what?
    Last edited by Ramiliez; 07-16-2013, 06:15 PM.

    Comment


    • #17
      Originally posted by Ramiliez View Post
      Its dog slow? i mean it takes 30s for journald to just page to end and i have only 1049322 lines there
      What command are you using? o.O

      Comment


      • #18
        Originally posted by Ericg View Post
        What command are you using? o.O
        journalctl

        Monster sized log files especially binary ones are not good for VMs how do you fix journald to not abuse disk space and still retain some history?
        Last edited by Ramiliez; 07-16-2013, 06:33 PM.

        Comment


        • #19
          Originally posted by Ramiliez View Post
          journalctl

          Monster sized log files especially binary ones are not good for VMs how do you fix journald to not abuse disk space and still retain some history?
          Ramiliez, this isn't directly directed at you its a comment about something ive noticed in general....

          Seriously people, some of the BEST linux documentation available comes from the Arch and Gentoo wikis. WE ARE YOUR TEST BED. I don't care if you are using Debian, Fedora, Ubuntu or whatever... if you have a question, check the Arch and Gentoo wikis. ONE OF THEM probably has an answer because a lot of other distros documentation is outdated or incomplete.

          Now, Ramiliez, to answer your question.

          From the arch wiki: https://wiki.archlinux.org/index.php/Systemd#Journal

          Journal size limit

          If the journal is persistent (non-volatile), its size limit is set to a default value of 10% of the size of the respective file system. For example, with /var/log/journal located on a 50 GiB root partition this would lead to 5 GiB of journal data. The maximum size of the persistent journal can be controlled by SystemMaxUse in /etc/systemd/journald.conf, so to limit it for example to 50 MiB uncomment and edit the corresponding line to:

          Code:
          SystemMaxUse=50M
          Refer to man journald.conf for more info.

          Comment


          • #20
            Originally posted by Ericg View Post
            Ramiliez, this isn't directly directed at you its a comment about something ive noticed in general....

            Seriously people, some of the BEST linux documentation available comes from the Arch and Gentoo wikis. WE ARE YOUR TEST BED. I don't care if you are using Debian, Fedora, Ubuntu or whatever... if you have a question, check the Arch and Gentoo wikis. ONE OF THEM probably has an answer because a lot of other distros documentation is outdated or incomplete.

            Now, Ramiliez, to answer your question.

            From the arch wiki: https://wiki.archlinux.org/index.php/Systemd#Journal

            ...or, you know, the man pages? journald.conf isn't exactly like the gcc entry

            Systemd related tools have phenomenal official documentation. I wish all projects were as them when it comes to docs.

            Comment


            • #21
              Originally posted by liam View Post
              ...or, you know, the man pages? journald.conf isn't exactly like the gcc entry

              Systemd related tools have phenomenal official documentation. I wish all projects were as them when it comes to docs.
              Wiki's are more likely to have specific examples, which is useful for when the documentation is unclear

              Comment


              • #22
                Originally posted by Ericg View Post
                Wiki's are more likely to have specific examples, which is useful for when the documentation is unclear
                Hey, I'm not against wikis, esp Arch/gentoo wikis, but the systemd man pages are just soooo good.
                I feel like I'm gushing a bit, but I'm just used to really poorly written "blueprints" which don't even document all the features for the given utility.

                Comment


                • #23
                  Originally posted by liam View Post
                  Hey, I'm not against wikis, esp Arch/gentoo wikis, but the systemd man pages are just soooo good.
                  I feel like I'm gushing a bit, but I'm just used to really poorly written "blueprints" which don't even document all the features for the given utility.
                  Oh I'm not disagreeing that systemd's man pages aren't good. But in my experience with man pages they are the exception, not the rule :/

                  Comment


                  • #24
                    So how can i backup journald logs?
                    How come that journal is 5.9M while it contains just 44 entries

                    [root@localhost journal]# du -h
                    5.9M ./fbb0c4896a9c0f680475c0a07edd2c5a
                    5.9M .

                    And btw reducing log size also removed a lot of logs but instead of keeping the newest logs im left garbage from 6 days

                    Comment


                    • #25
                      In my experience, if your log files are in a binary format then you are doing it wrong (TM).
                      Especially on Linux, you have just thrown all the traditional text processing tools (sed, grep etc) out the window.

                      There are ways to index text files for faster access etc. It's also easy to limit the size of the log file, or just create a new file whenever the size reaches a certain limit.

                      Now I will need a special tool just to read the log file? Microsoft went down that alley, and failed. Log files on Windows are horrible and inaccessible. Don't do that to Linux.

                      Comment


                      • #26
                        Originally posted by amehaye View Post
                        In my experience, if your log files are in a binary format then you are doing it wrong (TM).
                        Doing it wrong? Different is not wrong; there are some good arguments for binary logs, mostly related to security.

                        The only question here is what Fedora does by default, and honestly, who cares? Nothing is stopping you from using a different system logger, all messages are forwarded to a socket at /run/systemd/journal/syslog. You can tell journalctl not to store logs and have your syslog daemon listen on that socket, or you can run both. If you're unable to make these changes, chances are that you're unable to actually make use of the text logs, too.

                        Comment


                        • #27
                          Originally posted by Ramiliez View Post
                          So how can i backup journald logs?
                          How come that journal is 5.9M while it contains just 44 entries

                          [root@localhost journal]# du -h
                          5.9M ./fbb0c4896a9c0f680475c0a07edd2c5a
                          5.9M .

                          And btw reducing log size also removed a lot of logs but instead of keeping the newest logs im left garbage from 6 days
                          Your log file directory is named with a list of a thousand random hex chars? Gee... where have we seen this before?

                          Originally posted by amehaye View Post
                          In my experience, if your log files are in a binary format then you are doing it wrong (TM).
                          Exactly. I still remember leafing through MS's Log File Viewer just to find some half-helpful information. I'd prefer to forget about the experience altogether.

                          Comment


                          • #28
                            Originally posted by Ramiliez View Post
                            So how can i backup journald logs?
                            How come that journal is 5.9M while it contains just 44 entries

                            [root@localhost journal]# du -h
                            5.9M ./fbb0c4896a9c0f680475c0a07edd2c5a
                            5.9M .

                            And btw reducing log size also removed a lot of logs but instead of keeping the newest logs im left garbage from 6 days
                            Odd but regardless, checking the journald.conf man page also adds in this line

                            MaxRetentionSec=

                            0 is the default (which turns it off) Journal entries older than the specified time limit (in seconds) will be automatically deleted. You may append "year" "month" "week" "day" "h" or "m" to over ride the default time unit of seconds.

                            Comment


                            • #29
                              Originally posted by johnc View Post
                              Your log file directory is named with a list of a thousand random hex chars? Gee... where have we seen this before?



                              Exactly. I still remember leafing through MS's Log File Viewer just to find some half-helpful information. I'd prefer to forget about the experience altogether.
                              Remember 1 min shutdown time on Windows? they have it too and also there's another issue when you have entry in fstab Hard Driver for example and you physically unplugged it systemd will just refuses to let you log in

                              Comment


                              • #30
                                Originally posted by amehaye View Post
                                In my experience, if your log files are in a binary format then you are doing it wrong (TM).
                                Especially on Linux, you have just thrown all the traditional text processing tools (sed, grep etc) out the window.

                                There are ways to index text files for faster access etc. It's also easy to limit the size of the log file, or just create a new file whenever the size reaches a certain limit.

                                Now I will need a special tool just to read the log file? Microsoft went down that alley, and failed. Log files on Windows are horrible and inaccessible. Don't do that to Linux.
                                This special tool gives you text output in the standard format, which you can pipe to all the traditional text processing tools (hey, even to a text file! wow!)

                                Anyway, this has been lengthily debated, and journal is here to stay, as the main logging mechanism for fedora (and some other distributions).
                                The current debate is not whether journal is good, nor whether it should be used by default, nor whether other logging systems should or could be made incompatible.
                                It is about making it the only one installed by default, as opposed to (previously) one of two mechanisms installed by default.

                                Comment

                                Working...
                                X