This video was a life-saver. I've been dabbling in docker swarm for a while now but have been constraining services to single nodes due to lack of replicated storage. Thanks for the simple walkthrough, it made setup much less intimidating.
Followed both vids so far and they have been a HUGE help. We need a video now showing me how to install a docker or template via portainer ensure the Gluster volume is utilized.
Brother did you do the continuation from this? I am running my swarm with gluster, everything works but the volumes. It seems that is because of permitions.
Great video! I am going to keep playing with it and get used to creating and destroying volumes in GlusterFS. I also want to set it up with geo replication, so that's the next step, but the official documentation is so frustrating. Also, if you try to locate /test-volume … it does not exist in the original file system. It is in GlusterFS's own file system only, and that is what we are mounting to /mnt. That is the only way that it can make sense to me. And the link between the two file systems is that /glusterfs/volumes = /test-volume. But I might be very wrong here.
Great video I have everything setup and running. Im running a 12 pi cluster but I only see 23G that are being shared when each pi has 64G each. Can I add a SSD to this ?
In your next video are you going to cover the portainer plugin for glusterfs on raspberry pi? Here is a snippet to look into: docker plugin install --alias glusterfs mochoa/glusterfs-volume-plugin-aarch64 --grant-all-permissions In my case I was working with 64 bit pi400 with the 23.04 ubuntu server.
I'm new to linux, and my questions below may look immature. Sorry for that. 1) Can i use exisiting oracle block storage mounts as glousterfs volumes? 2) Does glousterfs nodes needs to be on the same cloud region or can this be from any cloud region as long as there is an active tunneling established acrossed the nodes? 3) What happens if one of the three nodes reboots or goes to a hung state, and will it affect the entire glusterfs system temporarily?
QQ. Do you recommend setting up each Ubuntu VMs, each with 2 separate disks, the first disk to run the OS, and the second as the GlusterFS replicated volume? Or could I provision the VMs to store OS and GlusterFS replicated volume configuration on one huge iSCSi LUN on my Proxmox hypervisor? Please let me know. Thanks.
Hey, great question! So, think of GlusterFS kind of like a smarter version of NFS. Instead of keeping all your data in just one spot, GlusterFS can spread it out or make copies across different machines in your swarm. This means you're not just pointing to where the data lives; you're actually sharing it around so that if one machine goes kaput, you're not left in the lurch. It's all about making sure your data is safe, sound, and readily accessible, no matter what happens. Hope that clears things up!
Thank you - all up and running on Bullseye 64bit. My only problem is after a restart the /mnt is not working. Master and nodes starting at the same time - so I think therfore I never get them mount over /etc/fstab - I am searching for a solution.
And... Where is the swarm setup? How to make 5 step guide is great, but mount local glusterfs, docker plug-in for this fs and other stuff is necessary. It is "zero level" tutorial 😢
Yeah, I'll be honest I was planning on creating more videos on swarm and making a series but it's run away from me a bit. This was to be part of a bigger implementation guide.
nearly at the end hit a snag... seeking to mount the GlusterFS volume to the /mnt. Got this exception. ERROR: Server name/volume name unspecified cannot proceed further.. Please specify correct format Usage: man 8 /usr/sbin/mount.glusterfs
@@Techdox I tried, but not thsi round. Perhaps it is a RPi issue, not sure. The share shows on each system but it ran into a road bump... needed to move on. Working on a startup self funded project so conserving time and resources pulled me off that till I get additional eyes on. I think it will need paired focus.
This video was a life-saver. I've been dabbling in docker swarm for a while now but have been constraining services to single nodes due to lack of replicated storage. Thanks for the simple walkthrough, it made setup much less intimidating.
Glad the video helped you!
Thanks for the great video. I'd been literally suffering with GFS until all went smooth with your walk through.
Best regards and wishes.
Big fan of docker swarm here, happy to see some refreshing videos about it.
Thank you :)
Followed both vids so far and they have been a HUGE help. We need a video now showing me how to install a docker or template via portainer ensure the Gluster volume is utilized.
Brother did you do the continuation from this? I am running my swarm with gluster, everything works but the volumes. It seems that is because of permitions.
Very interested in this project and your tutorials are very helpful, but I can't seem to find the next video in the series.
Been folliwing your great video tutorials, but did you ever do another video on using Docker Swarm after this one? cheer Matt
I did stop with the Swarm videos but do plan on making the series again
Great video! I am going to keep playing with it and get used to creating and destroying volumes in GlusterFS. I also want to set it up with geo replication, so that's the next step, but the official documentation is so frustrating.
Also, if you try to locate /test-volume … it does not exist in the original file system. It is in GlusterFS's own file system only, and that is what we are mounting to /mnt. That is the only way that it can make sense to me. And the link between the two file systems is that /glusterfs/volumes = /test-volume. But I might be very wrong here.
Looking for new GlusterFS content for months ;)
So is your manager node of the swarm the "host"? I've been lost the whole way through working on trying to get glusterfs to work.
The host is my server running Docker swarm, so each node runs Docker which means each node is a host
Great video I have everything setup and running. Im running a 12 pi cluster but I only see 23G that are being shared when each pi has 64G each. Can I add a SSD to this ?
Good stuff. Thanks
In your next video are you going to cover the portainer plugin for glusterfs on raspberry pi?
Here is a snippet to look into: docker plugin install --alias glusterfs mochoa/glusterfs-volume-plugin-aarch64 --grant-all-permissions
In my case I was working with 64 bit pi400 with the 23.04 ubuntu server.
I'm new to linux, and my questions below may look immature. Sorry for that.
1) Can i use exisiting oracle block storage mounts as glousterfs volumes?
2) Does glousterfs nodes needs to be on the same cloud region or can this be from any cloud region as long as there is an active tunneling established acrossed the nodes?
3) What happens if one of the three nodes reboots or goes to a hung state, and will it affect the entire glusterfs system temporarily?
How should you reference the glusterfs in docker compose for the swarm? Would this be correct or how should it be?
- /mnt/container-folder
QQ. Do you recommend setting up each Ubuntu VMs, each with 2 separate disks, the first disk to run the OS, and the second as the GlusterFS replicated volume? Or could I provision the VMs to store OS and GlusterFS replicated volume configuration on one huge iSCSi LUN on my Proxmox hypervisor? Please let me know. Thanks.
I personally would keep OS and Data seperated by disks
Can you mirror or raid mirror the shared volume storage for the swarm?
The swarm is looking for any storage you throw at it, so if you had a raid setup, you would point swarm to use that :)
Kind of confused a bit, is the data/volume replicated/copied across all nodes or just referencing their location?
Hey, great question! So, think of GlusterFS kind of like a smarter version of NFS. Instead of keeping all your data in just one spot, GlusterFS can spread it out or make copies across different machines in your swarm. This means you're not just pointing to where the data lives; you're actually sharing it around so that if one machine goes kaput, you're not left in the lurch. It's all about making sure your data is safe, sound, and readily accessible, no matter what happens. Hope that clears things up!
Thank you - all up and running on Bullseye 64bit. My only problem is after a restart the /mnt is not working. Master and nodes starting at the same time - so I think therfore I never get them mount over /etc/fstab - I am searching for a solution.
Glad this worked for you! In terms of your nodes, is there a reason they all come up at the same time?
And... Where is the swarm setup? How to make 5 step guide is great, but mount local glusterfs, docker plug-in for this fs and other stuff is necessary.
It is "zero level" tutorial 😢
Yeah, I'll be honest I was planning on creating more videos on swarm and making a series but it's run away from me a bit. This was to be part of a bigger implementation guide.
nearly at the end hit a snag... seeking to mount the GlusterFS volume to the /mnt. Got this exception.
ERROR: Server name/volume name unspecified cannot proceed further..
Please specify correct format
Usage:
man 8 /usr/sbin/mount.glusterfs
Hey did you end up getting this solved?
@@Techdox I tried, but not thsi round. Perhaps it is a RPi issue, not sure. The share shows on each system but it ran into a road bump... needed to move on. Working on a startup self funded project so conserving time and resources pulled me off that till I get additional eyes on. I think it will need paired focus.