Opened 9 years ago

Closed 9 years ago

#2646 closed enhancement (implemented)

Warn when less than 3 bandwidth scanners are running

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

Description

We should try harder to keep at least 3 bandwidth scanners running, or clients won't make use of the available bandwidth capacity.

See this graph from blutmagie for what happens when 2 or more of our bandwidth scanners are failing.

I'm going to implement a check in metrics-web to warn me when less than 3 directory authorities report measured bandwidth values in their votes.

Child Tickets

Attachments (1)

anonymizer2.blutmagie.de.tcpo-week.png (3.0 KB) - added by karsten 9 years ago.
Number of TCP connections to blutmagie

Download all attachments as: .zip

Change History (5)

comment:1 Changed 9 years ago by karsten

Summary: Warn when we less than 3 bandwidth scanners are runningWarn when less than 3 bandwidth scanners are running

comment:2 Changed 9 years ago by Sebastian

If maximum reliability is the goal here it seems the check should be "any bw auths fail"

comment:3 in reply to:  2 Changed 9 years ago by karsten

Status: newassigned

Replying to Sebastian:

If maximum reliability is the goal here it seems the check should be "any bw auths fail"

Right, or that.

Changed 9 years ago by karsten

Number of TCP connections to blutmagie

comment:4 Changed 9 years ago by karsten

Resolution: implemented
Status: assignedclosed

Implemented. The consensus-health checker sends me a warning whenever at least one of the bandwidth scanners fails, and the Nagios script contains a WARNING when 3 <= x < all bandwidth scanners are running and a CRITICAL message for x < 3. Closing.

Note: See TracTickets for help on using tickets.