Closed mckenziec closed 4 years ago
You need to up the open files limit (which is where the ulimit warning is coming from) - https://docs.riak.com/riak/kv/2.1.4/using/performance/open-files-limit/
Note that when you run riak console
after you have run riak start
it will try and start riak again (riak console
is just a way of starting riak when troubleshooting riak, it starts riak and outputs logs to the terminal rather than to file). So I think the riak console
may be failing as you already have riak started - and so it can't register the same listener twice - hence the "Protocol: ~tp: register/listen error: ~tp~n",["inet_tcp",epmd_close]}
error.
Before you start riak (using either start or console), make sure riak isn't already running on that node.
@mckenziec - I've not heard back, is all OK with this? Will close tomorrow if there is nothing further.
Sorry about the late reply martin. I realize you replied very quickly but things came up and distracted me, and I ended up moving on and haven't gone back to test your suggestion. I would suggest including the ulimit tuning in the Linux install instructions if its not merely a performance tweak, but really an installation requirement.
I will come back to riak and take another look though. My initial look was motivated by etcd's inability to function in a cluster where failure resulting in 1 node would prevent anyone from accessing kv data from it, because it goes into a RAFT election dead spiral. e.g. you need minimum 3 nodes to survive losing 1 node, but if it's reversed and 2 nodes go down that 1 node can't function at all. I ended up implementing my own limited kv forwarding service that doesn't attempt to deal with locking or value versioning. I just wanted to distribute node unique kv data. I know TMI. Thanks.
Version: 2.9.0p3
I was looking to give Riak KV a try and I haven't been able to start the server/console following the instructions here: https://docs.riak.com/riak/kv/2.2.3/setup/installing/rhel-centos.1.html https://docs.riak.com/riak/kv/2.2.3/setup/installing/verify/#starting-a-riak-node
Steps:
Start with a reasonable CentOS 7:
I have no clue what the problem might be. I have the crash dump attached.
erl_crash.dump.txt