On Fri 2013-06-28 18:25:38 -0400, Daniel Kahn Gillmor wrote: > sorry, i'm running out of time today, and need to head out. I was > unable to get a clean example of the problem to happen that would be > fixed by this proposed change -- i got tripped up by some other problems > that i haven't had the time to report.
ok, in trying to get this upgrade to work smoothly, i'm not convinced that the upgrade path is clean anyway. I set up a squeeze system, and did the following as root: apt-get install sks then, as debian-sks, i did: cd /var/lib/sks/dump wget http://ftp.prato.linux.it/pub/keyring/dump-latest/sks-dump-000{0,1}.php /usr/lib/sks/sks_build.sh cat > /var/lib/sks/DB/DB_CONFIG <<EOF set_mp_mmapsize 268435456 set_cachesize 0 134217728 1 set_flags DB_LOG_AUTOREMOVE set_lg_regionmax 1048576 set_lg_max 104857600 set_lg_bsize 2097152 set_lk_detect DB_LOCK_DEFAULT set_tmp_dir /tmp set_lock_timeout 1000 set_txn_timeout 1000 mutex_set_max 98304 EOF I looked up a user ID i could experiment with: gpg --list-packets 2>/dev/null < /var/lib/sks/dump/sks-dump-0000.pgp | head | grep ^:user I stored a reasonable fragment of this User ID the UID_FRAGMENT environment variable. and then as root again, i did: echo initstart=yes > /etc/defauilt/sks /etc/init.d/sks start This set up a server which i could check by searching for the user i found above: wget -O- "http://localhost:11371/pks/lookup?op=vindex&fingerprint=on&search=$UID_FRAGMENT"; so far, so good. but then the upgrade process fails, even with the new postinst script fixes: root@squeeze:~# dpkg --install sks_1.1.3-2+deb7u1_amd64.deb (Reading database ... 36497 files and directories currently installed.) Preparing to replace sks 1.1.1+dpkgv3-6+squeeze1 (using sks_1.1.3-2+deb7u1_amd64.deb) ... Stopping sks daemons: sksrecon.. sksdb.. done. Unpacking replacement sks ... Setting up sks (1.1.3-2+deb7u1) ... Installing new version of config file /etc/sks/sksconf ... Installing new version of config file /etc/sks/mailsync ... Installing new version of config file /etc/init.d/sks ... Installing new version of config file /etc/cron.daily/sks ... debian-sks uid check: ok debian-sks homedir check: ok db4.7_archive: DB_ENV->log_archive: DB_NOTFOUND: No matching key/data pair found dpkg: error processing sks (--install): subprocess installed post-installation script returned error exit status 1 Processing triggers for man-db ... Errors were encountered while processing: sks root@squeeze:~# a bit of digging online suggests that this error message might happen when one of the tables has no data in it (maybe a recon-related table, since this demo installation isn't gossiping with anyone?), so this might be another failed upgrade scenario that we need to take into account. argh. sorry to not have sorted this out yet. --dkg
pgptex9i8aEcl.pgp
Description: PGP signature
