# 2021-04-11-15:55:43+0800 tunning PD store-limit configuration for customer

- Cause: A customer scaled their cluster (id: 1379661944623214066) but the scaling process is too slow, we tune the PD parameters to accelerate this process
- Operation:

```
> kubectl exec -it -n tidb1379661944623214066 db-pd-0 -- sh
> cd /var/lib/pd-tls
> /pd-ctl --pd https://db-pd:2379 --cacert ca.crt --key tls.key --cert tls.crt store limit 1 200
> /pd-ctl --pd https://db-pd:2379 --cacert ca.crt --key tls.key --cert tls.crt store limit 10 200
> /pd-ctl --pd https://db-pd:2379 --cacert ca.crt --key tls.key --cert tls.crt store limit 7 200
> /pd-ctl --pd https://db-pd:2379 --cacert ca.crt --key tls.key --cert tls.crt store limit 4 200
> /pd-ctl --pd https://db-pd:2379 --cacert ca.crt --key tls.key --cert tls.crt store limit 5 200
> /pd-ctl --pd https://db-pd:2379 --cacert ca.crt --key tls.key --cert tls.crt store limit 11 200
> /pd-ctl --pd https://db-pd:2379 --cacert ca.crt --key tls.key --cert tls.crt config set replica-schedule-limit 1000
```

- Impact: The scaling process (migrate date regions) will be accelerated, correspondingly, the CPU usage of TiKV nodes will increase.
- Recovery operation: set the limit to the default value 15

```
> kubectl exec -it -n tidb1379661944623214066 db-pd-0 -- sh
> cd /var/lib/pd-tls
> /pd-ctl --pd https://db-pd:2379 --cacert ca.crt --key tls.key --cert tls.crt store limit 1 15
> /pd-ctl --pd https://db-pd:2379 --cacert ca.crt --key tls.key --cert tls.crt store limit 10 15
> /pd-ctl --pd https://db-pd:2379 --cacert ca.crt --key tls.key --cert tls.crt store limit 7 15
```
- Time window: 22:00:00 CST, July 15 ~ 22:00:00 July 17