Re: premature aborts?
Hi,
>>"Jason" == Jason Gunthorpe <jgg@gpu.srv.ualberta.ca> writes:
>> Also, does apt check secondary archives (I do not have any, at the
>> moment, but I shall rectify that)?
Jason> No, it checks only the first archived mentioned to have the
Jason> file. Is there a reason to support secondary archives?
Well, only in case the file could not be found or failed in
some other fashion (I sometimes get timemouts on some sites
intermittently -- I maen on some days they are good, on others they
are down).
Also, right now llug seems to be particularily bad --
__> find /var/spool/mirror/debian/ -name .in.\*
/var/spool/mirror/debian/hamm/hamm/binary-all/doc/.in.lg-issue11_2-3.deb.
/var/spool/mirror/debian/hamm/hamm/binary-all/doc/.in.lg-issue01to08_2-3.deb.
/var/spool/mirror/debian/hamm/non-free/binary-i386/games/.in.doom_1.10-16.deb.
The mirror fails on these files always at this point saying
I/O error -- and henve my local mirror is not in sync with the
Packages file, which mention these files.
In favour of general fault tolerance, maybe we should have a
final check when the packages are supposed to be on the local disk;
make sure the packages exist (stat), make sure the md5sum matches,
and then do the ordering step then.
This would fail if we stream download/unpack to minimize space
needed (and do not care about interactive time -- doewnloads are
interspersed with unpacks), but that is not how apt is implemented at
the time.
So, a final check of packages followed by a weeding out and
reordering process before unpacking would mean we upgraded the
maximal number of packages without falling over on our faces ;-)
In that case, we should only delete packages that actually
succeeded, maybe?
manoj
--
FIELD TESTED: manufacturer lacks test equipment.
Manoj Srivastava <srivasta@acm.org> <http://www.datasync.com/%7Esrivasta/>
Key C7261095 fingerprint = CB D9 F4 12 68 07 E4 05 CC 2D 27 12 1D F5 E8 6E
--
To UNSUBSCRIBE, email to deity-request@lists.debian.org
with a subject of "unsubscribe". Trouble? Contact listmaster@lists.debian.org
Reply to: