aboutsummaryrefslogtreecommitdiffstats
path: root/docs
Commit message (Collapse)AuthorAgeFilesLines
...
* offer again ungeneratedtv7 days1-1/+2
|
* stale delta offerstv7 days1-0/+3
|
* delta known to sendertv7 days1-1/+2
|
* delta propagationtv7 days1-1/+2
|
* benchmark bridge fast linktv7 days1-2/+3
|
* benchmark bridge slow linktv7 days1-1/+2
|
* benchmark bridgetv7 days1-1/+4
|
* one delta at a timetv7 days1-1/+5
|
* time blob messagestv7 days3-4/+4
|
* dashboard: hop delaystv7 days1-0/+1
|
* throughput from delivered messagetv7 days1-1/+1
|
* hop delaystv7 days2-0/+33
|
* benchmark members joiningtv7 days1-3/+3
|
* document messages that overtaketv7 days1-4/+20
|
* any message taken as deliveredtv7 days1-7/+9
|
* slow round back offtv7 days2-5/+10
|
* tree messages need no answertv7 days1-3/+10
|
* hello again when heard fromtv7 days1-1/+4
| | | | | | 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 slottv7 days1-3/+4
| | | | | | 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.
* no heal while both move ontv8 days2-2/+2
| | | | | | | The heal clock now also restarts when our heads and the peer's have both changed since the last STATUS that showed them differing. While writes go on, two neighbours's heads differ because of changes still on their way, and that alone no longer leads to a heal.
* pull after gracetv8 days1-1/+3
| | | | | | | | 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 lackingtv8 days2-3/+4
| | | | | | | 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 reachedtv8 days2-6/+11
|
* add member away benchmarktv8 days1-1/+2
|
* dashboard: show reach settv8 days1-0/+1
|
* dashboard: show writer treestv8 days2-0/+44
|
* update architecture doctv8 days1-4/+7
|
* writers from authorstv8 days1-5/+7
|
* reach set for the sessiontv8 days2-3/+5
|
* tree changes had alreadytv8 days1-1/+3
|
* announcement round counterstv8 days2-1/+7
|
* docs after treestv9 days1-1/+1
|
* announce what trees misstv9 days1-1/+6
|
* parallel propagation benchmarktv9 days1-1/+3
|
* quicker propagation benchmarktv9 days1-1/+2
|
* order members by distancetv9 days1-4/+6
|
* replace eager links by treestv9 days5-32/+13
|
* add writer treestv9 days1-0/+16
|
* add writer tree benchmarktv9 days1-1/+4
|
* measure 2-hop neighbourhoodtv9 days2-1/+20
|
* add propagation benchmarktv9 days1-0/+8
|
* dashboard: show eager links on the maptv9 days2-4/+9
|
* add eager link counterstv9 days2-1/+10
|
* introduce eager linkstv9 days2-1/+15
|
* skip neighbours with our headstv9 days2-2/+5
|
* announcements answered only if heads differtv9 days2-3/+4
|
* retry new peers before probingtv9 days1-1/+3
|
* probe list memberstv9 days1-1/+3
|
* links: add failovertv9 days1-0/+6
|
* swarm: links by whom we reachtv9 days1-1/+2
|