aboutsummaryrefslogtreecommitdiffstats
path: root/src/swarm
Commit message (Collapse)AuthorAgeFilesLines
...
* offer again ungeneratedtv7 days1-32/+53
|
* stale delta offerstv7 days1-0/+19
|
* blob window growstv7 days1-5/+10
|
* delta known to sendertv7 days1-1/+7
|
* delta propagationtv7 days2-2/+18
|
* one delta at a timetv7 days2-9/+53
|
* time blob messagestv7 days4-4/+46
|
* log why a deltatv7 days1-0/+6
|
* offered deltas lapsetv7 days1-4/+14
|
* throughput from delivered messagetv7 days2-2/+59
|
* hop delaystv7 days4-2/+128
|
* benchmark delays by pathtv7 days2-0/+15
|
* fix: change publisher stops with the runtimetv7 days1-3/+8
|
* benchmark what sync bytes aretv7 days2-3/+30
|
* changes from a peer never go backtv7 days1-0/+4
|
* older delivered messages leave us leveltv7 days1-1/+3
|
* benchmark where twice comes fromtv7 days2-4/+11
|
* resend only what is neededtv7 days1-1/+1
|
* benchmark delivered and twicetv7 days2-0/+30
|
* any message taken as deliveredtv7 days2-13/+23
|
* slow round back offtv7 days2-5/+86
|
* tree messages need no answertv7 days3-6/+40
|
* hello again when heard fromtv7 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 slottv8 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 thentv8 days2-4/+4
|
* pull after gracetv8 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 lackingtv8 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 reachedtv9 days2-14/+43
|
* links: only news pass ontv9 days1-6/+12
|
* dashboard: show writer treestv9 days1-26/+70
|
* writers from authorstv9 days3-27/+46
|
* reach set for the sessiontv9 days2-16/+10
|
* tree changes had alreadytv9 days1-4/+14
|
* announcement round counterstv9 days2-8/+23
|
* docs after treestv9 days1-5/+6
|
* announce what trees misstv9 days2-35/+115
|
* order members by distancetv9 days2-26/+59
|
* replace eager links by treestv9 days6-137/+35
|
* add writer treestv9 days5-27/+266
|
* add writer tree benchmarktv9 days1-0/+148
|
* measure 2-hop neighbourhoodtv9 days7-8/+136
|
* dashboard: show eager links on the maptv9 days1-5/+19
|
* add eager link counterstv9 days2-0/+21
|
* introduce eager linkstv9 days3-21/+100
|
* skip neighbours with our headstv9 days2-15/+37
|
* announcements answered only if heads differtv9 days4-7/+30
|
* retry new peers before probingtv9 days1-1/+1
|
* probe list memberstv9 days3-6/+17
|
* links: add failovertv9 days1-14/+129
|
* swarm: links by whom we reachtv9 days2-12/+25
|