Hi!
I am running a middle-man relay on a Raspberry Pi averaging 700 kBps. The board is very limited in processing power and memory. After some days, maybe a week, the Tor daemon runs out of memory and I have to start it manually again. I have tried different Tor versions, built alphas with and without the --enable-openbsd-malloc but it's no measurable difference; eventually a bad malloc will terminate the program.
To solve this I have added a cron job to restart the daemon on a daily basis. Is this an OK solution? Traffic-wise it takes around 30 minutes to get back to normal throughput but the alternative is long downtimes until I notice the crash. Will there be any problem with the directory interaction?
Relay info: Fast Guard Named Running Stable V2Dir Valid Tor 0.2.4.11-alpha
Kind Regards, peckman