Maybe so frequent restarts will increase the node and network failure rates (proportion of βtimeoutβ rounds), as many will interrupt normal round processing. Note that the βerrorβ status does not affect neither your nominal uptime nor your failure rate.
What would be the recommended timeframe for restarts?
By the way I notice in the gateway and node logs it seems that everything has stopped but in the dashboard a lot of SUCCESS and a couple of TIMEOUT rounds so the services must be doing something well, only the logging has stopped?
Well if I see the βupdating to ERRORβ message in the logs the node service is restarted by the xxwrapper script and service is continued with more rounds executed after that.
I describe the situation nothing is written to the logs but the server seems to work as the dashboard registers SUCCESS and some TIMEOUT rounds
I do not restart the services on node- or gatewayserver with cron
I guess we have to get the XX teams opinion on using cron, and if yes on what time schedule.
The reason why i would like to use it is so that i donβt have to reset the service everyday since my node gets stuck sometimes and shows offline in the dashboard. I talked with ben and keith about it. They say its not the node operators problem, but a fault in the software.
Sure you want to keep the services running and if cron is your only friend now, you have to stick to it. Hopefully the main reason for the problem will be found and solved.
From Keith aka LordVetinari in discord, on automatic service restarts:
βEvery time you kill your node mid-round you cause an error in two other nodes. FYI.β
βFor testing it really is better to let the node run its course even if it means running aground. Weβd much rather see unrecoverable errors than recoverable ones which are handled by the wrapper script.β
So the team discourages automatically restarting services with cron.