aws-hitcounter: remove temporary files as soon as possible
aws-hitcounter: ignore downloaded SRPM files
aws-hitcounter: decode special characters from URLs

count only hits from an appropriate CDN hostname
add option for infinite number of attempts to the hitcounter script
print more reasonable output from AWS hitcounter script

log every request that is sent to frontend

Don't use --keep-all-metadata
Search for comps.xml in chroot dir

Consolidate the two hitcounter scripts
Dump Resalloc ticket ID and hostname to backend.log
Automatically restart services in %post
Don't count RPMs downloaded from Mock
Attempt to sign multiple times
Try multiple attempts of creating GPG keys

fix slow priority queue filling
speedup the background-process spawner

copr_fix_gpg: automatically refresh CDN cache
copr_fix_gpg: don't use --skip-stat for copr-repo when RPMs are re-signed
action processor: properly return failures of the fork action
sign EL8+ with sha256 hash algorithm
copr_fix_gpg: add a new --chroot option
copr_fix_gpg: skip non-chroot directories
add hitcounter script for AWS CDN
backend: tasks in concurrent sandboxes reprioritized to be more fair

less aggressive final_prunerepo setting
analyze-results: dump data for projects' chroots
basic build tagging
better "regenerate repo" instructions
limit RubyGems and PyPI package names length
Disable coverage analysis during RPM build
python code for removing unused tarballs on dist-git server

Fixup ACR handling
Drop the unused pid file from the specfile

backend: use lock(timeout=5) to work-around fair-locks

request arch_noarch resalloc tag for source RPM builds
re-try /bin/sign call upon connection timeout
drop 'check_consecutive_build_fails' script

new weekly cron-job for analyzing storage use (graphs, statistics)
added some convenience globals into copr_backend.setup module
keep the max batch size really on the MAX_IN_BATCH limit
require up2date rpmbuild version
new action for removing CoprDirs (triggered by cron on frontend)
fix the CompsUpdate action
users now can disable appstream metadata generation themselves (without admins asistance)
handle results.json given by builder, and provide it to frontend (served as APIv3 later)
pruner: allow pruning also the finalized chroots on demand
invent FE-BE API version, so backend politely waits for an updated copr-frontend version

work with builders also over ipv6

fix logging traceback for module builds
call creatrepo_c with --update if possible
don't do full createrpeo with --rpms-to-remove

better fixes for the appstream-builder generated files
new helper script named /bin/copr-assure-permissions
do not rsync-copy permissions from the builder
log the partial "prunerepo" effects into resultdir
better, more verbose call_copr_repo logging

backend: make the walk_limited test not dependend on its output order, pt2