On 2019-12-11 11:48:00 [+0100], Christoph Berg wrote: > Re: Sebastian Andrzej Siewior 2019-12-07 > <20191207201131.v563o62fpmjnz7ol@flow> > > clamav can't migrate because the debci-test for pg-snakeoil fails. I > > *think* that the test itself is somehow borken since after installing > > the bytecode.cvd itself appears on my system. Could you please take a > > look? > > The test fails in my sid chroot as well because freshclam can't > download the database, /var/lib/clamav/ is empty except for a "tmp" > directory.
Do you have a special inet setup? Kind of web proxy or something like that. > $ cat /var/log/clamav/freshclam.log > Wed Dec 11 10:32:47 2019 -> -------------------------------------- > Wed Dec 11 10:32:47 2019 -> freshclam daemon 0.102.1 (OS: linux-gnu, ARCH: > x86_64, CPU: x86_64) > Wed Dec 11 10:32:47 2019 -> ClamAV update process started at Wed Dec 11 > 10:32:47 2019 > Wed Dec 11 10:32:47 2019 -> daily database available for download (remote > version: 25660) > Wed Dec 11 10:32:52 2019 -> WARNING: Mirror https://database.clamav.net is > not synchronized. > Wed Dec 11 10:32:52 2019 -> ERROR: Unexpected error when attempting to update > database: daily > Wed Dec 11 10:32:52 2019 -> WARNING: fc_update_databases: fc_update_database > failed: Up-to-date (1) > Wed Dec 11 10:32:52 2019 -> ERROR: Database update process failed: Up-to-date > (1) > Wed Dec 11 10:32:52 2019 -> ERROR: Update failed. > Wed Dec 11 10:32:52 2019 -> -------------------------------------- Could you start `freshclam' by hand with --verbose (not sure if --debug works) and provide more output? It appears that the version downloaded is one less than available and that is where things go south. > > However. The complete database (/var/lib/clamav/) contains today 454MiB. > > I don't know what your self-test but using > > > > | clamav$ cat unit_tests/input/clamav.hdb > > | aa15bcf478d165efd2065190eb473bcb:544:ClamAV-Test-File > > > > as a databse will find the un-offical test file: > > > > |$ clamscan -d clamav.hdb split.clam.exe > > |split.clam.exe: ClamAV-Test-File.UNOFFICIAL FOUND > > Using a smaller database instead of downloading the whole thing for > each test run makes sense. > > We just need to figure out how to compile it to a > /var/lib/clamav/*.cld database that cl_load() understands. At the > moment the database path is not configurable in pg_snakeoil, maybe we > need to change that as well. Bug. The .cld files are signed by upstream if I am not mistaken. I will try to check. > Christoph Sebastian