Opened 3 years ago

Closed 3 years ago

#20922 closed defect (implemented)

Consider faster retry schedule for primary guards

Reported by: nickm Owned by:
Priority: Medium Milestone: Tor: 0.3.0.x-final
Component: Core Tor/Tor Version:
Severity: Normal Keywords:
Cc: asn, chelseakomlo Actual Points:
Parent ID: #19877 Points:
Reviewer: Sponsor:

Description

30 minutes may be too low. Should we aim for 10?

Child Tickets

Change History (5)

comment:1 Changed 3 years ago by nickm

2767ae18064a87 on prop271_030_v1 has a tweak here.

comment:2 Changed 3 years ago by nickm

Cc: asn chelseakomlo added
Status: newneeds_review

comment:3 Changed 3 years ago by nickm

Parent ID: #20822#19877

comment:4 Changed 3 years ago by asn

I think this is a good change for the cases where we marked our primaries as down while we were offline, and now we are only dealing with confirmed guards. This change will make us go back to our primaries much faster.

That said, this change has performance implications when the top primary guards are legitimately down, since we will have to wait about 2 minutes for each primary guard connection to timeout every 10 minutes.

I currently think that this change is worth doing, but I would need to test this branch more in real life situations to see if the performance penalty is too annoying.

comment:5 Changed 3 years ago by nickm

Resolution: implemented
Status: needs_reviewclosed

Calling this implemented since the branch is merged, but we'll have to keep an eye on this.

Note: See TracTickets for help on using tickets.