feed-care

Repository to keep alive feeds

Directory contains necessary scripts that if run from crontab, can make sure that a system running price feed using setzer, and geth --light or parity --light will become more reliable.

Prerequisities

• mail command must be operational to send email to arbitrary address.
• setzer must be installed
• seth must be installed
• geth must be installed (if you install parity, the scripts wont restart it)
• hostname command must be installed (package hostname under ubuntu)

Installation

Lets assume the git clone base directory is DIR.

0. Edit alert.conf and update EMAIL, and ACC variables. (See Setup settings for details!)
1. Install all the packages under Prerequisities.
2. mkdir /root/.scripts # create dir for scripts
3. cp $DIR/\*.sh /root/.scripts # copy all script files from DIR to /root/.scripts
4. cp setzer-bot.service /etc/systemd/system/ && systemctl daemon-reload && systemctl enable setzer-bot && systemctl start setzer-bot # create setzer-bot service (this one automatically restarts setzer if it stops), and starts at boot time automatically

If you use geth as Ethereum client as well

5. cp UTC--xxxx /root/.ethereum/keystore/ #Add your ethereum feed account keystore to geth client.
6. Add and edit the file called /root/.scripts/p.txt and enter your ethereum feed account password.
7. cp geth.service /etc/systemd/system/ && systemctl daemon-reload && systemctl enable geth && systemctl start geth #create service for geth (this one automatically restarts geth if process gets terminated), and starts at boot time automatically
8. Use crontab_example to edit crontab file crontab -e

If you use parity as Ethereum client as well

9. cp UTC--xxxx /root/.local/share/io.parity.ethereum/keys/ethereum/ #Add your ethereum feed account keystore to geth client.

10. Add and edit the file called /root/.scripts/p.txt and enter your ethereum feed account password.

11. cp parity.service /etc/systemd/system/ && systemctl daemon-reload && systemctl enable parity && systemctl start parity #create service for parity (this one automatically restarts parity if process gets terminated), and starts at boot time automatically

12. Use crontab_example to edit crontab file crontab -e

13. You are all set. :)

Note: You can install both geth and parity clients at the same time.

Setup settings

/etc/setzer.conf

#allowed maximum time the scripts wait before rendering node unavailable
export RPC_TIMEOUT=50s

Note: in previous implementations $ALERT_NODE_TIMEOUT was used. But in current implementation we just use setzer's $RPC_TIMEOUT for the same purpose.

alert.conf

PLEASE EDIT THIS FILE BEFORE USING SCRIPTS

#email for alerts
export EMAIL="YOUR EMAIL ADDRESS"

#ethereum account for paying feed
export ACC="YOUR FEED ACCOUNT ADDRESS"

#balance threshold in Ether under which alert is sent
export THRESHOLD=0.01

#disk space in percentage to send alert
export DISK_SPACE_THRESHOLD=90

#parity rpc port number (should be different from GETH_RPC_PORT)
export PARITY_RPC_PORT=8545

#parity node port number, where parity expects outher nodes
#should be different from GETH_NODE_PORT
export PARITY_NODE_PORT=30303

#geth rpc port
export GETH_RPC_PORT=8555

#geth node port number, where geth expects outher nodes
#should be different from PARITY_NODE_PORT
export GETH_NODE_PORT=30305

Alert if balance too low

alert_balance.sh

Alerts email address if balance of feeds account is below THRESHOLD Ether defined in alert.conf.

Script should be run every hour.

Alert if geth stops working

alert_geth.sh

Alerts user if geth client stops updating blocks (ie. geth is not operational). Script should be run every hour. (Note: should not be called more often, because it takes geth sometimes 40 minutes to even start syncing blocks.)

Alert if parity stops working

alert_parity.sh

Alerts user if parity client stops updating blocks (ie. parity is not operational).
Script should be run every hour.

Alert if file system is close to full

alert_filesystem.sh

Alerts EMAIL if free disk space is under DISK_SPACE_THRESHOLD in alert.conf.
This script should be run once every hour.

Alert if feed was not updated for more than 6 hours

alert_txcount.sh

This script alerts EMAIL if feed update count did not increase. (Meaning the feed has not been updated since last run of this script.) This script should be run in every 6 hours.

Alert if there is only one ethereum node left to connect to

alert_client.sh

This script alerts EMAIL if there is only 1 ethereum client to connect to. This script should be run in every 30 minutes.