Memory leak. InfluxDB not correctly work

__Environment info:__

Linux 3.10.0-1127.19.1.el7.x86_64 x86_64
InfluxDB v1.8.3 (git: 1.8 563e6c3d1a7a2790763c6289501095dbec19244e)
VM KVM Scientific Linux release 7.9 (Nitrogen)
2 Core, 4Gb RAM, 90 Gb HDD
SWAP 18 GB!!! USED 13 GB!!!!
telegraf 1.17
Grafana 6.3.6

__Config:__
influxdb.conf https://gist.github.com/Scrik/17742d9d63b32ecf60f8150974a97af9

__Performance:__

We received a bunch of alerts from Grafan and when I went to see what happened, it turned out that the telegraph could not write to the Influx. Influx ate all the RAM and died. I added SWAP and tried to do something with it, to which it ate 13 GB of SWAP. Now it has become easier, but the dosikhpor telegraph cannot record the data.
[diagnostics-1611418352.txt](https://github.com/influxdata/influxdb/files/5860743/diagnostics-1611418352.txt)
[goroutine-1611418624.txt](https://github.com/influxdata/influxdb/files/5860744/goroutine-1611418624.txt)
[heap-1611418779.txt](https://github.com/influxdata/influxdb/files/5860747/heap-1611418779.txt)
[iostat.txt](https://github.com/influxdata/influxdb/files/5860748/iostat.txt)
[shards-1611417906.txt](https://github.com/influxdata/influxdb/files/5860749/shards-1611417906.txt)
[stats-1611417933.txt](https://github.com/influxdata/influxdb/files/5860750/stats-1611417933.txt)

Database SIZE
58G telegraf
119M _internal
648K backups
576K reports
340K munin

Start with the attached diagnostics, goroutine, heap, iostat, shards, and stats reports, using the stated InfluxDB, Telegraf, Grafana, and Linux environment as context. Determine what consumes memory and prevents writes; done means the leak is identified and the reported failure is no longer reproducible under the same workload.

Written by the indexing model from the issue text.