Bug 1033956 - journal logs are huge
Summary: journal logs are huge
Keywords:
Status: CLOSED NOTABUG
Alias: None
Product: Fedora
Classification: Fedora
Component: systemd
Version: 19
Hardware: Unspecified
OS: Unspecified
unspecified
unspecified
Target Milestone: ---
Assignee: systemd-maint
QA Contact: Fedora Extras Quality Assurance
URL:
Whiteboard:
Depends On:
Blocks:
TreeView+ depends on / blocked
 
Reported: 2013-11-24 18:49 UTC by Mohammed Arafa
Modified: 2013-12-13 00:34 UTC (History)
7 users (show)

Fixed In Version:
Doc Type: Bug Fix
Doc Text:
Clone Of:
Environment:
Last Closed: 2013-12-13 00:34:07 UTC
Type: Bug


Attachments (Terms of Use)

Description Mohammed Arafa 2013-11-24 18:49:28 UTC
Description of problem:

this is not the first time and i have seen these logs reach into GIGABYTES in size.
i delete them and then on the next reboot they can be 7mb already. i suspect this is becuase they are binary logs and not text files. this is a sample from this morning. note the directory is already 33mb after the reboot. and 2 users logged on.

marafa@notebook:/var/log/journal$ du -sh *
33M     fed72e1673be4583a85b470cfbec51ef
marafa.org:/var/log/journal$ du -sh */*
26M     fed72e1673be4583a85b470cfbec51ef/system.journal
3.6M    fed72e1673be4583a85b470cfbec51ef/user-1000.journal
3.6M    fed72e1673be4583a85b470cfbec51ef/user-1003.journal


Version-Release number of selected component (if applicable):


How reproducible:
every SINGLE time

Steps to Reproduce:
1. rm -rf /var/log/journal/*
2. reboot
3. du -sh /var/log/journal/*

the large size of the log file of gb i suspect is due to a lot of journal activity and/or a large time fram

Actual results:


Expected results:


Additional info:

Comment 1 Zbigniew Jędrzejewski-Szmek 2013-11-28 06:39:06 UTC
The size of individual files and also the total disk usage scale proportionally to file system size and the amount of free space, up to 4GB. So it's entirely expected for the journal to grow to gigabytes in size, space permitting.

Journal files are created with a few fixed-size tables, with their size based on the maximum file size. So again, 7 mb journal file when it is created is not unexpected. This space is reserved for future use, and should be filled as entries are stored in this file. You can view some information about the file with 'journalctl --header'.

Can you provide the output from 'journalctl --header' and 'journalctl|wc' ?

Comment 2 Mohammed Arafa 2013-11-29 13:25:47 UTC
marafa@notebook:~/tmp$ journalctl --header
File Path: /var/log/journal/fed72e1673be4583a85b470cfbec51ef/system.journal
File ID: bc85f8594e47481796c832b6c1cb31d7
Machine ID: fed72e1673be4583a85b470cfbec51ef
Boot ID: fc20d92bba0f4cf8b6a35188265a6bce
Sequential Number ID: fc20d92bba0f4cf8b6a35188265a6bce
State: ONLINE
Compatible Flags:
Incompatible Flags: COMPRESSED
Header size: 240
Arena size: 6147856
Data Hash Table Size: 233016
Field Hash Table Size: 333
Rotate Suggested: no
Head Sequential Number: 1
Tail Sequential Number: 73123
Head Realtime Timestamp: Thu 2013-11-28 15:07:57 EST
Tail Realtime Timestamp: Fri 2013-11-29 08:23:18 EST
Objects: 10493
Entry Objects: 3204
Data Objects: 5581
Data Hash Table Fill: 2.4%
Field Objects: 45
Field Hash Table Fill: 13.5%
Tag Objects: 0
Entry Array Objects: 1661
Disk usage: 5.8M

File Path: /var/log/journal/fed72e1673be4583a85b470cfbec51ef/user-1000.journal
File ID: 8bdeed111c26489db9aded24f0c8aa72
Machine ID: fed72e1673be4583a85b470cfbec51ef
Boot ID: fc20d92bba0f4cf8b6a35188265a6bce
Sequential Number ID: fc20d92bba0f4cf8b6a35188265a6bce
State: OFFLINE
Compatible Flags:
Incompatible Flags: COMPRESSED
Header size: 240
Arena size: 3792656
Data Hash Table Size: 233016
Field Hash Table Size: 333
Rotate Suggested: no
Head Sequential Number: 2125
Tail Sequential Number: 73045
Head Realtime Timestamp: Wed 2013-11-20 13:12:47 EST
Tail Realtime Timestamp: Fri 2013-11-29 08:11:16 EST
Objects: 397
Entry Objects: 44
Data Objects: 187
Data Hash Table Fill: 0.1%
Field Objects: 24
Field Hash Table Fill: 7.2%
Tag Objects: 0
Entry Array Objects: 140
Disk usage: 3.6M

File Path: /var/log/journal/fed72e1673be4583a85b470cfbec51ef/user-1003.journal
File ID: 3ea9213787f1497cb7fce9e072b75917
Machine ID: fed72e1673be4583a85b470cfbec51ef
Boot ID: fc20d92bba0f4cf8b6a35188265a6bce
Sequential Number ID: fc20d92bba0f4cf8b6a35188265a6bce
State: OFFLINE
Compatible Flags:
Incompatible Flags: COMPRESSED
Header size: 240
Arena size: 3763984
Data Hash Table Size: 233016
Field Hash Table Size: 333
Rotate Suggested: no
Head Sequential Number: 18573
Tail Sequential Number: 72965
Head Realtime Timestamp: Sat 2013-11-23 10:00:14 EST
Tail Realtime Timestamp: Fri 2013-11-29 08:10:06 EST
Objects: 222
Entry Objects: 21
Data Objects: 106
Data Hash Table Fill: 0.0%
Field Objects: 21
Field Hash Table Fill: 6.3%
Tag Objects: 0
Entry Array Objects: 72
Disk usage: 3.5M

File Path: /var/log/journal/fed72e1673be4583a85b470cfbec51ef/system~
File ID: a84eeb8ebdc44b65a6fb086a064fd846
Machine ID: fed72e1673be4583a85b470cfbec51ef
Boot ID: 093aa14175154192b8a71d5c047a3a05
Sequential Number ID: 093aa14175154192b8a71d5c047a3a05
State: OFFLINE
Compatible Flags:
Incompatible Flags: COMPRESSED
Header size: 240
Arena size: 51859216
Data Hash Table Size: 233016
Field Hash Table Size: 333
Rotate Suggested: no
Head Sequential Number: 1
Tail Sequential Number: 74907
Head Realtime Timestamp: Wed 2013-11-20 13:02:16 EST
Tail Realtime Timestamp: Thu 2013-11-28 14:32:18 EST
Objects: 167825
Entry Objects: 74851
Data Objects: 84195
Data Hash Table Fill: 36.1%
Field Objects: 48
Field Hash Table Fill: 14.4%
Tag Objects: 0
Entry Array Objects: 8729
Disk usage: 49.5M




marafa@notebook:~/tmp$ journalctl|wc
  78125  956482 9626199

Comment 3 Zbigniew Jędrzejewski-Szmek 2013-11-29 20:56:42 UTC
61MB / 78125 records = 780 bytes / record.

This is not particularly efficient, but since the journal includes two timestamps and additional information for each record, it's not outrageous either.

Also please note that if you allow those files to grow, storage should become more efficient as time goes on, because duplicate fields are not stored multiple times, so at some point the average number of bytes / record should go down.

Comment 4 Lennart Poettering 2013-12-13 00:34:07 UTC
We log substantially more data, this hence necessarily is larger thatn the old text files. I don't see anything to fix here really...


Note You need to log in before you can comment on or make changes to this bug.