[Date Prev][Date Next]   [Thread Prev][Thread Next]   [Thread Index] [Date Index] [Author Index]

Re: --repackage error?

Valery Reznic wrote:
--- Hal Wine <hal@reactivity.com> wrote:

I'm hoping to use --repackage to allow rollback of
our product. I had no problems with using that option during development
and initial testing, but a corrupt RPM was generated on the first use in
the QA lab (of course!)

When reinstalling the error message is:
warning: Expected size: 43696763 = lead(96)+sigs(180)+pad(4)+data(43696483)
warning: Actual size: 43697108
approuter ####################error: unpacking of archive
failed on file /usr/local/spam/eggs;3f823a90: cpio:
MD5 sum mismatch

from a command line of:
rpm  -Uvh --force --nomd5 --nodigest --noscripts
command exit code was 6

As near as I can tell, the repackaged rpm got
generated with the correct md5 digest for 'eggs', but the actual file is
corrupted (different length). I can do a "rpm2cpio {package} | cpio -id
*/eggs" and get no error messages, but the extracted file is, indeed,

- I haven't seen this behavior reported on the list.
Is it common?

- Since this corruption apparently happened at
--repackage time, is there a way to verify integrity before erasing?

- I'm currently checking out rpmrebuild -- does
anyone have serious flight time with this package?

During development rpmrebuild was tested on whole linux distributions and only few packages was not rebuilded. You can see rpmrebuild web to see list of these packages and reasons.


Yes, I've seen that. fwiw, it turns out that in my case it is likely the corruption occurred before the "rpm --repackage" was run.

As I understand rpmrebuild, it also would have built the rpm using the existing corrupted file. It wouldn't have helped in this case.

So, I've decided not to use rpmrebuild at this juncture, as that just means more software I have to install and support, and I'm basically lazy :)


[Date Prev][Date Next]   [Thread Prev][Thread Next]   [Thread Index] [Date Index] [Author Index]