I am running IPFS cluster on my local machine using multiple docker containers by scripting a yaml file. I have successfully run a 22 cluster node cluster. When i try to add nore nodes, they exeit with code 1, I want to extend my network to 50 nodes. Is there anyone who can guide me to extend my network? I can use cloud platform to run another cluster using docker, but how can i merge both clusters?
What is the error? What do the logs say? Is your local machine running out of memory?
I have managed the storage space. I have changed the default storage of 10GB to 2GB for each of the ipfs node by editing the configuration file of each of the nodes
ipfs-cluster invokes su-exec https://github.com/ipfs-cluster/ipfs-cluster/blob/6289b56db4b20cebd97bb94ef0ac1a37574d853f/docker/start-daemons.sh#L12 while we removed it from Kubo 3 months ago refactor(ci): simplify Dockerfile and add docker image testing (#10021) · ipfs/kubo@c5868a8 · GitHub
# Using gosu (~2MB) instead of su-exec (~20KB) because it's easier to # install on Debian. Useful links: # - https://github.com/ncopa/su-exec#why-reinvent-gosu # - https://github.com/tianon/gosu/issues/52#issuecomment-441946745
And the ipfs-cluster docker image is based on Kubo’s one https://github.com/ipfs-cluster/ipfs-cluster/blob/6289b56db4b20cebd97bb94ef0ac1a37574d853f/Dockerfile-bundle#L22
On one hand, sorry silent breaking change. On the other hand the way docker let chain constructions very far like this looks extremely fragile to me and expose way too much internals rather than segmenting things in their own robust islands like containers promised to do.
Which docker container are you using though?
The bundle is just for testing amd @jorropo is right: it’s broken. We don’t distribute builds from this Dockerfile though.
1 Like
Thank you for your reply @Jorropo @hector.
It is quite a high-level explanation. I am sorry. It is not clear to me. I am a student and a new learner. I am trying to use IPFS as an underlying storage layer. This is the yaml file that I used to create ipfs cluster. I have attached the system architecture to make it clear (just mentioned 6 nodes, but I need 50). According to the architecture, I am exposing 3 apis (in yaml file, i exposed apis at 1,5, 11) at 1,4 and 6 to cope with a single point of failure. Could you please check my attached yaml file and suggest some changes that i need to scale up my system? For less than 20 nodes, this yaml file works fine, when I configure more nodes in yaml file and try to run, some of the nodes exit with code 1(as mentioned above).
Dear @hector , I have rebuilt the image with gosu. Now, when I try to run more than 15 cluster nodes, the cluster peer exits with code 1.**
Here are the logs for cluster 17. This issue arises only when I try to run more than 15 cluster peers. it works fine with less than 15 nodes.
As I said in DM, this is not a problem with cluster, this is a problem with your machine being hammered by running too many things on it.