hyperledger fabric (application) is creating too many files in one directory
Our deployment of hyperledger fabric ( application ) is on 1.1 cluster has ubuntu based 5 worker nodes which has mounted to nfs to the same machine (nfs machine is rhel based os vm and shares the folder via nfs to multiple nodes), hyperledger chain height is more than 10 Million.
Now we have observed that if the peer (application) crashes in order to get it up again it takes a lot of time even more then 2 hours. On closer observation we have found that "ledgersData/chains/index/" has around more then 5000+ files. Even doing a simple ls takes a lot of time.
Is that the reason for the slowness of the application while waking up?
Has somebody else encountered similar problem as well?
Is there a problem with filesystem or application
The files are stored on a GPFS based NFS and nfs is mounted on the multiple containers that are writing.
Any pointers in the right directions are appreciated.