Red Hat Bugzilla – Bug 1029604
Boot takes 27 seconds longer with /var/log/journal than without
Last modified: 2014-06-17 20:36:03 EDT
I believe this one affects RHEL7 too - judging from bug #1006386 comment #29 I believe this is systemd-journald design flaw
I managed to break my testing installation (beyond my ability to repair, bug #1029527) so I cannot provide exact numbers now, but if I recall correctly, the boot took 20 - 25 seconds with several services blamed to take more than ten or twelve seconds
after doing a fresh install, the machine booted in 12.5 seconds
the testing procedure consisted of writing random things to logs for some tenths of hours (with pauses in between to workaround rate limiting and to become a bit closer to the real world example where I saw this problem on Fedora)
note that simply "mkdir /var/log/journal ; reboot" is enough to get data written to it, systemd-journald runs by default
also note that these numbers are from a virtual machine run on a host with SSD and plenty of RAM, real hardware could be much slower (if I get it right that it accesses many 4 KiB blocks on disk)
+++ This bug was initially created as a clone of Bug #1006386 +++
Following patches were backported to rhel7 systemd repo, will be in next update.
fbb6341 journald: mention how long we needed to flush to /var in the logs performance
eda4b58 journal: simplify pre-allocation logic performance
248c78c journal: allow journal_file_copy_entry() to work on non-local files performance
e5462cd journal: fix iteration when we go backwards from the beginning of an array chain element performance
f268980 journal: optimize bisection logic a bit by caching the last position performance
a676e66 journal: when appending to journal file, allocate larger blocks at once performance
This request was resolved in Red Hat Enterprise Linux 7.0.
Contact your manager or support representative in case you have further questions about the request.