UNIX for Advanced & Expert Users

Dear All,

I'm using AWS EC2 instance for my application. My application is high disk I/O based and EFS could not be used in my case.

So, i need to build my own NFS server on Ec2 instance. I'm looking for High availability solution for my disk which i shared for NFS. Looking for builtin snapshot solution and 99.9% availablity. Is there any software raid or any solution which you can recommend to me?

Thanks
Bala

Can you clarify, do you need two boxes with NFS as clustered resource or ?
For snapshots and stuff (outside cloud provider), ZFS filesystem should offer the same, if used on NFS server as filesystem.
There is also an options of LVM snapshots.

You can use it on open solaris implementations such as OmniOS or modern linux with openzfs implementation.
OpenZFS

This is what is proven and working.

There is BRTFS as well as a GPL COW filesystem, with snapshots, which i cannot suggest, since i'm not sure if it is mature enough for production usage.

Regards
Peasant.

Thank you Peasant.

I'm looking for

1. Built-in snapshotsolution (Daily, and weekly)
2. Yes. Two Clustered NFS server. So that if one NFS server goes down, i can bring another one.

Please suggest me

Thanks
Bala

Will those 2 NFS boxes have same disk presented or data replication is required sided with NFS ?
What does high IO mean, are you running databases or ?

You do understand that NFS is not a clustered file system.
So writing in same file from multiple nodes will not work as you expect.

Can you elaborate a bit more about the architecture and application and what it does ?
Perhaps glusterfs, HDFS or ISCSI could be a better advice.

Regards
Peasant.

Peasant - I'm trying to setup GitLab instance on AWS EC2. It requires shared file system such as NFS. This application will have high I/O transaction.

The GitLab vendor doesn't support AWS EFS file system due to its limitation by nature. So, i want to build custom NFS server which will be shared on 4 servers.

So, I'm looking for HA solution to my NFS share.

1. Built in sanpshot solution for accidental file deletion and curruption
2. Since the NFS share disk going to reside on EBS volume, i want to ensure the disk will be safe.
3. Looking for solution to replicate the EBS volume data to another disk on the same machine or different machine.

Thanks
Bala

------ Post updated at 03:23 PM ------

Peasant - I'm trying to setup GitLab instance on AWS EC2. It requires shared file system such as NFS. This application will have high I/O transaction.

The GitLab vendor doesn't support AWS EFS file system due to its limitation by nature. So, i want to build custom NFS server which will be shared on 4 servers.

So, I'm looking for HA solution to my NFS share.

1. Built in sanpshot solution for accidental file deletion and curruption
2. Since the NFS share disk going to reside on EBS volume, i want to ensure the disk will be safe.
3. Looking for solution to replicate the EBS volume data to another disk on the same machine or different machine.

Thanks
Bala

I'm not sure what suggest here, since i have not worked with clouds so much.

But having replicated disks of any kind in cloud sounds bad.
As you depend on network latency for any kind of disk replication, your iops will suffer.
Similar as they suffer now using cloud builtin distributed disks, EFS.

Can you use just one gitlab instance, and amazon tools to backup / snapshot / copy the data around ?

If you said metal, things would be different since you control every aspect of implementation.

Regards
Peasant.