[tahoe-lafs-trac-stream] [tahoe-lafs] #1280: if bucket_counter.state or lease_checker.state can't be written, stop the node with an error message

[tahoe-lafs]

#1280: if bucket_counter.state or lease_checker.state can't be written, stop the
node with an error message
--------------------------------+--------------------------------
     Reporter:  francois        |      Owner:  zooko
         Type:  defect          |     Status:  reopened
     Priority:  major           |  Milestone:  1.9.0
    Component:  code-nodeadmin  |    Version:  1.8.1
   Resolution:                  |   Keywords:  pickle reliability
Launchpad Bug:                  |
--------------------------------+--------------------------------

Comment (by zooko):

 Replying to [comment:7 davidsarah]:
 > Replying to [comment:6 zooko]:
 > > Why stop the node? Because there is no other reliable way to get the
 operator's attention.
 >
 > Seems to me that's a bug right there.

 Well, let's open a ticket for it, but I still think that even if we fix
 that issue by providing a reliable way to get the operator's attention,
 that stopping the node is still the right thing to do when we discover
 something wrong like this, because (a) it increases the likelihood of
 getting the operator's attention, (b) it reduces the chance of damage or
 exploitation, and (c) it helps the operator with post-mortem diagnostics.

 > There should be some other way to get the operator's attention, and if
 there were, there would be lots of things that could use it without
 introducing SPoFs by stopping the node.

 What do you mean ''S''!PoFs?

 > Straw-man suggestion: send (rate-limited) email to an admin address
 giving a link to the log.

 I guess the [source:trunk/docs/logging.rst#incident-gatherer Incident
 Gatherer] is perfect for this.

-- 
Ticket URL: <http://tahoe-lafs.org/trac/tahoe-lafs/ticket/1280#comment:8>
tahoe-lafs <http://tahoe-lafs.org>
secure decentralized storage