UTP Interop

[njgheorghita]

More of a comment than an actual issue... But I think this could be of value to those of us who are debugging this library atm.

I wanted to confirm my suspicion as to what was causing the difference in performance when running glados against fluffy / trin. For context, running glados against fluffy resulted in drastically improved performance when compared to running glados against trin. It appeared as though fluffy was having no trouble with utp transfers, while trin was struggling. Since trin-bridge is the only bridge running, the conclusion was that utp txs for trin -> fluffy were working just fine, whereas trin -> trin was broken (for bodies & receipts ... aka any significantly long utp tx).

Results from running trin-bridge in backfill mode for 15mins. trin-bridge is run in isolation against a single client. Each utp tx is logged and marked as success & failure.

trin-bridge -> trin

trin-bridge -> fluffy

For whatever reason, it appears as though trin -> fluffy utp txs work flawlessly. Whereas trin -> trin utp txs are problematic. Hopefully this can help shed some insight as to where the bugs might reside.