Ensure that NTP is properly set up on all nodes. Particularly for Cassandra DO NOT use anything else other than ntp. Here are some helpful blogs that explain why:
How can I see if NTP is correctly set up?
This is an important part of your setup, particularly for your Cassandra nodes. You should use
ntpd and our ansible scripts to ensure it is installed correctly - but you can still check it manually if you prefer. The following 2 sub-sections explain both approaches.
I used your ansible scripts and prefer to have automated checks
Then the easiest way is to use this ansible playbook
I am not using ansible and like to SSH into hosts and checking things manually
The following shows how to check for existing servers connected to (assumes
ntpq is installed)
which should yield something like this:
remote refid st t when poll reach delay offset jitter ============================================================================== time.example. .POOL. 16 p - 64 0 0.000 0.000 0.000 +<IP_ADDR_1> <IP_ADDR_N> 2 u 498 512 377 0.759 0.039 0.081 *<IP_ADDR_2> <IP_ADDR_N> 2 u 412 512 377 1.251 -0.670 0.063
if your output shows _ONLY_ the entry with a
refid and a lot of 0s, something is probably wrong, i.e.:
remote refid st t when poll reach delay offset jitter ============================================================================== time.example. .POOL. 16 p - 64 0 0.000 0.000 0.000
What should you do if this is the case? Ensure that
ntp is installed and that the servers in the pool (typically at
/etc/ntp.conf) are reachable.