switch to persistent peerstore

[raulk]

Perhaps this helps:

[NumDhts: 40, Uptime: 5m0s, Memory Usage: 57 MB, TotalPeers: 737/1213, Total Provs: 0, BootstrapsDone: 3]
[NumDhts: 40, Uptime: 5m5s, Memory Usage: 67 MB, TotalPeers: 736/1225, Total Provs: 0, BootstrapsDone: 32]
[NumDhts: 40, Uptime: 5m10s, Memory Usage: 77 MB, TotalPeers: 751/1239, Total Provs: 0, BootstrapsDone: 32]
[NumDhts: 40, Uptime: 5m15s, Memory Usage: 82 MB, TotalPeers: 745/1251, Total Provs: 0, BootstrapsDone: 32]
[NumDhts: 40, Uptime: 5m20s, Memory Usage: 87 MB, TotalPeers: 729/1260, Total Provs: 0, BootstrapsDone: 32]
[NumDhts: 40, Uptime: 5m25s, Memory Usage: 90 MB, TotalPeers: 707/1267, Total Provs: 0, BootstrapsDone: 32]
[NumDhts: 40, Uptime: 5m30s, Memory Usage: 50 MB, TotalPeers: 706/1280, Total Provs: 0, BootstrapsDone: 32]
[NumDhts: 40, Uptime: 5m35s, Memory Usage: 54 MB, TotalPeers: 681/1285, Total Provs: 0, BootstrapsDone: 32]
[NumDhts: 40, Uptime: 5m40s, Memory Usage: 58 MB, TotalPeers: 663/1290, Total Provs: 0, BootstrapsDone: 32]
[NumDhts: 40, Uptime: 5m45s, Memory Usage: 62 MB, TotalPeers: 646/1299, Total Provs: 0, BootstrapsDone: 32]
[NumDhts: 40, Uptime: 5m50s, Memory Usage: 67 MB, TotalPeers: 639/1309, Total Provs: 0, BootstrapsDone: 32]
[NumDhts: 40, Uptime: 5m55s, Memory Usage: 71 MB, TotalPeers: 633/1322, Total Provs: 0, BootstrapsDone: 32]
[NumDhts: 40, Uptime: 6m0s, Memory Usage: 77 MB, TotalPeers: 616/1333, Total Provs: 0, BootstrapsDone: 32]
[NumDhts: 40, Uptime: 6m5s, Memory Usage: 81 MB, TotalPeers: 602/1335, Total Provs: 0, BootstrapsDone: 32]
[NumDhts: 40, Uptime: 6m10s, Memory Usage: 84 MB, TotalPeers: 583/1342, Total Provs: 0, BootstrapsDone: 32]

[Kubuxu]

Wont this increase our allocation velocity?

[raulk]

Yes but in a shared setup, the cache is useless because it'll just be thrashing all the time.

[Kubuxu]

Depending on size and type of cache.

[Kubuxu]

We should have metrics regarding the hit rate. Unfortunately, open census was disabled because it was allocating too much (Why knows more).