`tools-harbor-1` is crashing often because Redis takes up all the available memory.
The alert "Project tools instance tools-harbor-1 is down" has fired multiple times in the past week.
`top` shows redis-server is using 93% or RAM, load average is very high, and `kswapd` is using 18% CPU, but `free` doesn't show any swap being used.
```
top - 10:25:23 up 11 min, 2 users, load average: 51.97, 42.77, 23.21
Tasks: 189 total, 2 running, 187 sleeping, 0 stopped, 0 zombie
%Cpu(s): 2.8 us, 6.8 sy, 0.0 ni, 0.4 id, 89.9 wa, 0.0 hi, 0.0 si, 0.0 st
MiB Mem : 16010.4 total, 150.9 free, 15808.9 used, 50.7 buff/cache
MiB Swap: 0.0 total, 0.0 free, 0.0 used. 10.4 avail Mem
PID USER PR NI VIRT RES SHR S %CPU %MEM TIME+ COMMAND
587 root 20 0 1781280 18956 0 S 24.8 0.1 0:25.26 contain+
93 root 20 0 0 0 0 R 18.4 0.0 4:28.16 kswapd0
580 prometh+ 20 0 1753576 35632 0 D 6.2 0.2 0:39.25 prometh+
3759 10000 20 0 1372824 72652 0 S 5.8 0.4 1:59.40 harbor_+
1512 10000 20 0 730296 9732 0 S 5.6 0.1 0:22.18 harbor_+
4903 10000 20 0 1674324 359736 0 D 5.6 2.2 1:57.93 harbor_+
637 root 20 0 3310224 65940 0 S 4.7 0.4 0:36.64 dockerd
5436 root 20 0 17364 3856 504 D 1.5 0.0 0:03.17 htop
1522 10000 20 0 729856 5888 0 S 1.3 0.0 0:07.76 registr+
2341 root 20 0 1303092 8796 0 S 1.1 0.1 0:06.91 prometh+
1 root 20 0 164376 2984 0 D 0.9 0.0 0:06.78 systemd
280 root 20 0 64764 1076 4 D 0.9 0.0 0:05.66 systemd+
4872 root 20 0 1452068 4528 0 S 0.9 0.0 0:06.84 contain+
1760 root 20 0 1453220 5452 0 S 0.8 0.0 0:01.11 contain+
1835 999 20 0 15.7g 14.6g 0 S 0.8 93.4 1:41.24 redis-s+
5468 root 20 0 13032 4100 600 D 0.8 0.0 0:01.90 wmf-aut+
5484 root 20 0 17616 368 28 D 0.8 0.0 0:00.24 exim4
root@tools-harbor-1:~# free
total used free shared buff/cache available
Mem: 16394604 16186616 155984 1164 52004 11976
Swap: 0 0 0
```