Hello,

As sometimes happens, I notice my nodes getting into swap heavily.

This time:

------------ KiB Mem: 1010576 total, 897388 used, 113188 free, 5144 buffers KiB Swap: 1048572 total, 123592 used, 924980 free. 61732 cached Mem

PID USER PR NI VIRT RES SHR S %CPU %MEM TIME+ COMMAND 4740 debian-+ 25 5 565448 259032 15204 R 63.2 25.6 1621:47 tor 4765 debian-+ 25 5 491636 200812 13136 S 32.9 19.9 678:36.65 tor

total used free shared buffers cached Mem: 1010576 870392 140184 4 1144 64604 -/+ buffers/cache: 804644 205932 Swap: 1048572 123592 924980 ------------

After killing PID 4740 and starting that instance again (now PID 25456):

------------ KiB Mem: 1010576 total, 691264 used, 319312 free, 4336 buffers KiB Swap: 1048572 total, 58624 used, 989948 free. 209692 cached Mem Unknown command - try 'h' for help PID USER PR NI VIRT RES SHR S %CPU %MEM TIME+ COMMAND 4765 debian-+ 25 5 491636 195900 8584 R 38.3 19.4 678:56.29 tor 25456 debian-+ 20 0 441092 240824 140452 S 29.9 23.8 0:28.48 tor

total used free shared buffers cached Mem: 1010576 663816 346760 12 4344 209692 -/+ buffers/cache: 449780 560796 Swap: 1048572 58624 989948 ------------

How come killing a Tor instance which was shown to be using 259 MB of RAM, has freed up 355 MB, even counting the new copy of Tor which was started in its place and is already using 240 MB? The math only adds up to 19 MB delta, not 355.

As this memory usage doesn't seem to be attributed to Tor, it reminds me of that major memory leak which was fixed some time ago, so maybe it was not fixed entirely?