When a user's computer (not a relay) having established Tor circuits goes down into operating system defined states of deep sleep (hibernates) - to be restarted (awoken) at a later time - is state maintained in the net relative to those circuits, both at the Tor circuit protocol level and as underlying TCP connections ?

What are the timeouts if any applicable to such suspended Tor connections ?

When the user's computer later is restarted from hibernation, is the logical link with the current "guard" restored or recreated , OR will Tor stall, and need to be restarted - or at least, a newnym signal must be issued ?

1 Answer 1


Gone with the wind =) The hibernation resume is on your local point of network only, all the rest of the nodes are wide awake, so circuits are gone because of connections are dropped. A newnym signal is rarelyin order in this kind of situations: my Windows laptop slept and after 10 minutes I brought him back awake - Tor have told all the intimates of his relationship with that idea in the error log but was able to recover itself in 2-3 minutes

  • So, looking at the question in reverse so to say - from the point of vue of the guard relay, what is needed "on the wire" for it to "know" the user's Tor is still on line and circuits should be kept ? Are periodic "Tor pings", sort of, exchanged on the wire ? i mean, when a user has Tor running but is not using it for a while, circuits do NOT time out by themselves, or do they ?
    – Noino O
    Feb 21, 2017 at 18:33
  • it's a bit more simply: it's handled on TCP keepalives level: when the other side of connection sleeps - it just stops responding to them, and it crushes on the TCP pipe level.
    – Alexey Vesnin
    Feb 21, 2017 at 19:03

You must log in to answer this question.

Not the answer you're looking for? Browse other questions tagged .