hubble
A public mirror for the whole atmosphere hubble.microcosm.blue
Star
Fork
Watch

Commits on posts-count
partition the big scan
committed
lil total-by-collection counter
committed
wat wat
committed
audit counts
committed
bump eat-rocks (faster incremental restore) backup
committed
fix: don't create the checkpoint dir, rocks does
committed
configurable initial backup delay
committed
initial eat-rocks push
committed
bump deps
committed
bump a few deps
committed
bump versions for release
committed
oops skip slots
committed
Merge commit 'edcab0b0ddf79ab5e2e93be0cac8f346307f1afb'
committed
out-of-scope reevaluation: scan on startup
committed
debug: where are we losing the commits
committed
store `prev` from the commit object
committed
configurable archive export compression
committed
update todos
committed
turn off open admin mod writes for now lol
committed
Merge branch 'bridgy-auth-getrepo-offload'
committed
wire in the deep crawl
committed
deep crawl
committed
show did instead of index on error bridgy-auth-getrepo-offload
committed
handle response encoding (compression) in hubble
committed
conditionally install tokio-console subscriber
committed
buffer more output when writing archives
committed
track in-flight over response streaming
committed
getRepo bench
committed
split limit waits from request times
committed
opt-in auth token forward-across-redirect thing
committed
tolerate missing tooBig on #commit
committed
configs for server binds + upstream limits
committed
add hubble request metrics
committed
fix test name
committed
send an auth token with getRepo (optionally)
committed
resync queue guages
committed
get more timing info for resyncs
committed
blehhhhhh
committed
tokio-console
committed
Revert "please let this not work"
committed
please let this not work
committed
i guess it's this poll timeout...
committed
one more micro-optimization
committed
micro-optmize mushrooms
committed
maybe 16?
committed
moar mushroom requests at a time
committed
batch-prefetch queued resyncs
committed
bypass relay for getRepo on first-seen
committed
ok chill on the relay + allow more open files
committed
bump relay concurrency
committed
evict more actors at a time
committed
tests for record did interning
committed
use interned dids for records
committed
big machine bit faster
committed
one memtable per flush
committed
bleh. plz faster.
committed
go faster
committed
metrics for resynced bytes by ingest/batch-put
committed
shutdown: don't wait on compactions, log
committed
downgrade file-remove fail notice (it's normal)
committed