MEV boost has this option:
-request-timeout-getheader int
timeout for getHeader requests to the relay [ms] (default 950)
This value (950 ms) is chosen to be slightly less than the BUILDER_PROPOSAL_DELAY_TOLERANCE=1 sec spec value to allow some network latency between the beacon node and MEV boost.
However, I often see the duration field in the GET /eth/v1/builder/header/... log higher than that:
Examples with v1.11
INFO[XXXT00:25:12.121Z] http: GET /eth/v1/builder/header/... 200 duration=0.982362 method=GET path=/eth/v1/builder/header/... status=200 version=v1.11.0
And even with v1.12, > 1 second:
level=info msg="http: GET /eth/v1/builder/header/... 200" duration=1.010395 method=GET path=/eth/v1/builder/header/... status=200 version=1.12
The consequence of that is, even if MEV boost already has some relays bid, the connected beacon node falls back on the locally built block after the BUILDER_PROPOSAL_DELAY_TOLERANCE cutoff.
MEV boost has this option:
This value (950 ms) is chosen to be slightly less than the
BUILDER_PROPOSAL_DELAY_TOLERANCE=1 secspec value to allow some network latency between the beacon node and MEV boost.However, I often see the
durationfield in theGET /eth/v1/builder/header/...log higher than that:Examples with
v1.11And even with
v1.12, > 1 second:The consequence of that is, even if MEV boost already has some relays bid, the connected beacon node falls back on the locally built block after the
BUILDER_PROPOSAL_DELAY_TOLERANCEcutoff.