Couple of points.
1. The truncation test only works if the I/O buffer size is 8192 and no
more than one buffer has been written. So its still possible that a
file could have been truncated - and still pass the truncation test.
I'm not sure there is a simple way to fix this test. It is most likely
true though that CS.cfg will be larger than 8192 bytes, and this is the
default buffer size, so it *might* be OK to keep this test ..
2. I think time stamped config backups should occur automatically,
unless explicitly disabled. That means that for old servers, where the
config parameter is not set, the default bahavior should default to true
(and not false). There are a number of existing instances out there
that would benefit from this fix.
Ade
On Thu, 2014-06-19 at 22:31 -0700, Matthew Harmsen wrote:
After a spirited discussion on the IRC channel, I have restored the
archival functionality and re-factored the following bug to account
for items in the discussion.
Please review the attached patch which address the following bug:
* Bugzilla Bug #1061442 - RFE - ipa-server should keep backup of
CS.cfg
Note that this bug is ONLY relevant to the RHEL 6 platform.
The actual bug contains new documentation, a revised sample SRPM which
contains the patch and a spec file, and the comments in the bug have
been updated to provide ADDITIONAL TESTING instructions.
It would be great to have ACKs for this by early Friday, 06/20/2014,
so that I can generate official builds of 'pki-core'.
Thanks,
-- Matt
P. S. - Although the use of check sums was discussed during the IRC
session, I utilized the 'cmp' command to perform byte by byte
comparisons
instead since the 'cmp' command will supposedly stop as soon
as any difference is detected (thereby reducing the time needed to
read
the entire contents of both files).
_______________________________________________
Pki-devel mailing list
Pki-devel(a)redhat.com
https://www.redhat.com/mailman/listinfo/pki-devel