aboutsummaryrefslogtreecommitdiffstats
path: root/src/swarm
Commit message (Collapse)AuthorAgeFilesLines
* unreached sources hand overtv3 days1-1/+26
|
* chunk timeout adaptstv3 days1-6/+74
|
* manifests asked at oncetv3 days1-45/+57
|
* measure deflated chunkstv3 days2-8/+28
|
* deflate blob chunkstv3 days2-5/+29
|
* deflate blob chunkstv3 days2-5/+44
|
* epoch tag in synctv3 days2-9/+34
|
* no swarm ID in synctv3 days2-24/+9
|
* announce only to those behindtv3 days2-12/+31
|
* measure sync envelopestv3 days2-0/+24
|
* dashboard: swarm-wide epoch shinetv3 days1-8/+31
|
* asked known elsewheretv3 days3-2/+23
|
* predecessors go alongtv3 days1-3/+18
|
* covered neighbours have ittv3 days3-16/+54
|
* new changes per messagetv3 days2-1/+10
|
* neigbours by transporttv3 days1-1/+21
|
* own tree shapetv3 days1-10/+136
|
* drop full mesh stattv3 days1-3/+7
|
* delta from held headstv3 days1-1/+4
|
* offer again ungeneratedtv4 days1-32/+53
|
* stale delta offerstv4 days1-0/+19
|
* blob window growstv4 days1-5/+10
|
* delta known to sendertv4 days1-1/+7
|
* delta propagationtv4 days2-2/+18
|
* one delta at a timetv4 days2-9/+53
|
* time blob messagestv4 days4-4/+46
|
* log why a deltatv4 days1-0/+6
|
* offered deltas lapsetv4 days1-4/+14
|
* throughput from delivered messagetv4 days2-2/+59
|
* hop delaystv4 days4-2/+128
|
* benchmark delays by pathtv4 days2-0/+15
|
* fix: change publisher stops with the runtimetv4 days1-3/+8
|
* benchmark what sync bytes aretv4 days2-3/+30
|
* changes from a peer never go backtv4 days1-0/+4
|
* older delivered messages leave us leveltv4 days1-1/+3
|
* benchmark where twice comes fromtv4 days2-4/+11
|
* resend only what is neededtv4 days1-1/+1
|
* benchmark delivered and twicetv4 days2-0/+30
|
* any message taken as deliveredtv4 days2-13/+23
|
* slow round back offtv4 days2-5/+86
|
* tree messages need no answertv4 days3-6/+40
|
* hello again when heard fromtv4 days2-2/+19
| | | | | | While we wait for a peer's HELLO_REPLY, anything else that comes from it makes our HELLO go again at once, at most once a second, instead of when the backoff says.
* unanswered pull frees the slottv4 days2-1/+23
| | | | | | When the answer to a pull is overdue (the sync reply timeout: 1 s in the simtests, 10 s in production), the pull gives up the slot, and the next waiting neighbour is pulled at once.
* overdue sync asked after from thentv4 days2-4/+4
|
* pull after gracetv4 days2-4/+33
| | | | | | | | Before pulling, a node waits one measured round trip to the announcer (1 s at most) and pulls only what is still missing. If the round trip isn't known yet, it pulls at once, as now. On the clean benchmark links the round trip is a few milliseconds, so those scenarios shouldn't change.
* announcements answered when lackingtv4 days4-10/+34
| | | | | | | A neighbour now answers an announcement only if it lacks some of the announced heads. Heads that merely differ are no longer enough, since they do while writes go on. The check reads the document, so it runs off the message consumer, as the pull check already does.
* unanswered members not reachedtv5 days2-14/+43
|
* links: only news pass ontv5 days1-6/+12
|
* dashboard: show writer treestv5 days1-26/+70
|
* writers from authorstv5 days3-27/+46
|