14.4.6.21 seemed like it started out using less memory than what 15.x did. My usual time for things to go bad averaged about 25-30 days on 14.4.6.21. 15.x started out for me using a bit more memory from the beginning. Right now, at basically 20 days, my 15.x eth6 is using 20.2MB.tzr916 wrote:I'm curious too about the ETH6 random reboots. I think the longest I've gone is about 60 something days. Never kept an eye on the memory. Right now mine is at 20 days and 18692kB (still on v14.4.6.21).
Factors that influence the amount of time for things to go south might include the use of a TA (and/or which brand), number of computers/devices that are accessing it, number of channels pushed out by the cable system, etc etc. Lots of variables. The one consistent thing is just that the memory usages climbs fairly steadily, at least for me, for every firmware I've tried since 13.x.
Since I can't see inside the eth6's running OS, there is no way to know what is chewing up all the memory over time, if its a memory leak, a normal Linux caching more stuff over time issue, etc.
But hopefully Ceton will take these reports of memory usage increasing over time, see if it correlates to why the devices eventually get slow or crash/reboot, and hopefully eventually fix them. Its a bit frustrating to find shows that do not record every few weeks because the eth6 took a dump on itself and rebooted, or stopped communicating properly with the TA. I avoid a lot of that now by pre-emptively rebooting things if I notice channel change times slowing, but spontaneous reboots have sometimes happened to me after a day or two of uptime, so there are no guarantees.