Page MenuHomePhabricator

certspotter on einsteinium has issues talking to external
Closed, DeclinedPublic


we are getting cronspam from einsteinium with content like:

/usr/bin/certspotter: 2017/04/04 09:02:31 Error retrieving STH from log: Get net/http: request canceled while waiting for connection
/usr/bin/certspotter: 2017/04/04 20:01:56 Error retrieving STH from log: Get dial tcp i/o timeout
/usr/bin/certspotter: 2017/04/05 05:02:34 Error fetching consistency proof between 1088952 and 1090451 (if this error persists, it should be construed as misbehavior by the log): Get net/http: timeout awaiting response headers

Event Timeline

Dzahn created this task.Apr 5 2017, 11:54 PM
Restricted Application added a subscriber: Aklapper. · View Herald TranscriptApr 5 2017, 11:54 PM
Dzahn added a comment.Apr 6 2017, 11:11 PM

I can't reproduce it when manually running the command. Looks like intermittent on the remote side:

root@einsteinium:/etc# sudo -u certspotter /usr/bin/certspotter -watchlist /etc/certspotter/watchlist -state_dir /var/lib/certspotter/state

faidon moved this task from Inbox to Up next on the observability board.Jul 12 2017, 5:25 PM
faidon claimed this task.Jul 24 2017, 3:22 PM
faidon moved this task from Up next to In progress on the observability board.
faidon closed this task as Declined.Jul 24 2017, 3:39 PM

This is basically an artifact of the CT logs failing to respond every now and then, which certspotter complains about. It doesn't happen often.

We can do 2>/dev/null on the cronjob, but that runs the danger of certspotter being completely broken and us never hearing about it. Ideally, we'd have a way to only output errors if they have appeared more than N times/X days, but given how rudimentary the check is right now (just a cronjob), it's not going to be easy.

I'm inclined to decline this -- if it becomes a larger problem, we can always revisit. Let me know if you disagree :)