Opened 4 years ago

Closed 19 months ago

#15809 closed defect (invalid)

onionoo instances disagree on observed and advertised bandwidth

Reported by: cypherpunks Owned by:
Priority: Medium Milestone:
Component: Metrics/Onionoo Version:
Severity: Keywords:
Cc: Actual Points:
Parent ID: Points:
Reviewer: Sponsor:

Description

This is an _example_ for relay 4AA98421045E123101DBFB811B141349739E4B4C ('relays_published' = 2015-04-25 08:00:00)

tpo's instance:

"observed_bandwidth":37601326,

the cthulhu's instance:

"observed_bandwidth":40158236

https://atlas.torproject.org/#details/4AA98421045E123101DBFB811B141349739E4B4C
https://atlas.thecthulhu.com/#details/4AA98421045E123101DBFB811B141349739E4B4C

Child Tickets

Change History (3)

comment:1 Changed 4 years ago by cypherpunks

a more extreme example:
relay: 0A61BBC3C8E0114B96A7A3FD80A58A6818BEDE1A

observed_bandwidth

17933247 vs. 0

comment:2 Changed 4 years ago by karsten

Here's what happens: onionoo.thecthulhu.com is usually faster at updating than onionoo.torproject.org, and updating details documents happens for a period of 5 to 15 minutes in the background before the relays_published timestamp is updated. As a result, the same request to torproject.org might have changed 10 or 20 minutes later even with the same relays_published timestamp. If we want to avoid this, we'll have to switch to a database and update all documents in a single transaction. The current design simply stores documents in the file system, which doesn't support such a thing. I'm aware that this defeats the purpose of the If-Modified-Since header.

comment:3 Changed 19 months ago by iwakeh

Resolution: invalid
Status: newclosed

Closing, b/c the cthulhu is gone and there are now Tor operated instances rotating.

Note: See TracTickets for help on using tickets.