Configuring a Synology HA Cluster and Showing What Happens When You Pull The Plug!

[@LAWRENCESYSTEMS]

Lawrence Systems Synology Tutorials lawrence.technology/synology/ Synology Models that support HA kb.synology.com/en-me/DSM/tutorial/Can_I_create_a_High_Availability_cluster_with_two_different_Synology_servers ⏱ Timestamps ⏱ 00:00 Synology HA Tutorial 00:50 Synology HA Requirements 01:32 Demo Lab Setup 04:28 HA Cluster Setup 06:36 HA Cluster Failover Switching 10:42 HA Hard Failover Pulled Plug

[@Noodles.FreeUkraine]

Nice video, thanks Tom. It's a bit of a rabbit hole, though, isn't it? Raid is not a backup, it just increases availability. Now we have a redundant raid storage, which also just increases availability. So we need another one of those NAS to store a backup, since any data corruption would also replicate straight to the other NAS. And then we still don't have an off-location backup, NAS no. 4. 🙈 I wonder how the big guys handle all those double, triple and more data needs and still sell the GB for pennies on the dollar. Sure looks like a humungous dollar sink. 😬

[@jmugoose30]

I have a small business and use Synology with HA. I have a raid in my Synology which is set up solely for the backup that is different than the drives which are shared for storage. So the backups are stored on multiple hard drives and on both servers. Then yes I do have a separate Synology that is smaller that is another backup for the “live” data drives. It is a lot of drives and NAS hardware but still fairly cheap for the redundancy. I figure if a drive fails I have single NAS repairability. If a. NAS fails (which I have had happen) I am still up and running with backups. I semi regularly restore files from a backup just to verify (test files not a full restore) from both the HA cluster and the secondary NAS. Still much cheaper than losing data. Mine are 8 bay Synology loaded with 10GB seagate drives so all in im still in the $10,000 range for secure sleeping at night

[@ondrejstanek2157]

Can I use hyper backup, snapshot replication within HA cluster ? If active server fails will replication, snapshoting work on passive server ?

[@marco.lop88]

Hi Tom, Greate ilustration ! It would be interesting if you simulate an outage on the Alpha Server.

[@LAWRENCESYSTEMS]

I did a simulation of the server set as active in the video

[@marco.lop88]

Saw it now, thank you 👍

[@christopherdesouza7339]

Pretty good video. If you have the means I actually recommend using one of the 1GB ports for heartbeat. It is supported and heartbeat requires very little bandwidth. Instead of using 1 switch I use 2 switches. One 10GB port goes to each switch from each NAS and they are teamed on each NAS. This way even if I loose a switch I am still still up. All my HyperVisor devices are the same. Also gives me a chance to verify no affect for switch changes or firmware updates as I can do one and verify up. Plus switch maintenance does not take NAS offline.

[@nickharvey5149]

Tom which DAC cables were you using please?

[@LampJustin]

Can someone share some insights to how Synology is doing their ha? Are they using keepalived, drbd, Gluster, nfs-ganesha, pacemaker?

[@ilanbirman]

Hello, thank you very much for your video, I have 1 Synology FS3017 in production that only has LUNs configured. I just bought a second FS3017 after 2 years since the equipment is so critical that I need to leave it in high availability. The disks that were purchased are of the same capacity but different brand and model. Is there a problem with that? Thank you

[@jeffreyooi1971]

Hi, do you need to assign IP addresses to the heartbeat connection for the 2 NAS?

[@xWhiteKidHD]

This is like RAID 110

[@jasonchi9336]

What is the procedure to replace or upgrade drives?

[@d00dEEE]

Tom already mentioned the redundant switch issue, but what if the heartbeat connection fails? Does the heartbeat function then fail over to the switch connections? Do these boxes allow for a pair of redundant heartbeat connections? (Yeah, I know, a DAC is extremely unlikely to fail, but I'm just curious from an academic perspective.)

[@LAWRENCESYSTEMS]

They do have a way to setup path redundancy for the heartbeat connection by using link aggregation to provide failover redundancy and load balancing. This feature does not require a switch between the connections.

[@d00dEEE]

@@LAWRENCESYSTEMS Question answered, thanks, Tom!

[@loclhst]

Have you ever ran into any issues with iscsi restarting? My company ran the Synology HA a couple (maybe 5?) Years ago on dsm5, and anytime there was a data sync between the two units, it would either pause or restart iscsi which caused lots of problems for the 3 VM hosts we had connected. We a sfp+ 10g copper link between the two and gigE copper uplink to the switch for the servers. Caused so many issues that we ended up getting a different storage solution and sold the Synology units.

[@LAWRENCESYSTEMS]

Not a problem I have run into, but we are running the most recent version.

[@shotalezhava625]

hello, I have error The system partition layouts of the following drives on both server are incompatible:Drive1, Drive2, Drive3. I have RAID 10, Capacity and inserted slots are similar.

[@densonngumo7028]

Can one have a HA setup with TrueNas/FreeNas?

[@LampJustin]

Not really, yet. You can have an almost ha with ZFS replication, so you'd loose like 5min at most, but not real time replication. TrueNas scale tries to do that though. It will be using Gluster to achieve it. Imho if you want real ha, not like Synology and others, go Ceph ;) It's actually really easy to setup up with docker!

[@LampJustin]

And if scale is using nfs-ganesha then their solution is pretty good.

[@Darkk6969]

@@LampJustin CEPH is great but with one caveat. You need three nodes for it to work properly. You *can* run CEPH with just two nodes and use something like raspberry pi as a witness server but it's not recommended.

[@jeffwads]

Great time to be alive and in IT, ain’t it.

[@CapitaineWEB-FR]

I still wonder how to upgrade from 6.x to 7.x when using HA. 😬 Using our Synology as a datastore in XCP-NG.

[@LAWRENCESYSTEMS]

It's easy and if you already have the system in HA just run the updater and it will walk you through the steps.

[@popeter]

whats prob the best method for replication to a secondary site with a smaller NAS? e.g. i have a DS1621+ but want to have a DS220+ at another site for important files that users at site can also access the same files that are replicated across the 2 NAS's

[@Sertsch]

Rsync, syno drive sync or the new c2 service

[@vgamesx1]

Syncthing is excellent, but as with a lot of sync tools you have to be careful how you set it up, otherwise it could delete things.

[@mithubopensourcelab482]

On the face of it, Synology offers excellent feature. When we compare with TrueNas, this feature is missing. But hold on, if you have enabled snapshots and getting replicated another TrueNas Server, you are mostly done. The flexibility TrueNas offers that you can select the dataset's which needs replication, you have all the flexibility in the world when the syncing should be done. Another point is that you simply requires TrueNas on the other end, and while you are replicating, you can use replication server with other datasets of its own, making both TrueNas server are active at the same time. You can even replicate snapshots of both the servers simultaneously on one another [ both way ] . However, you can not create [ at least as of now ] one HA IP you will never get 100 % HA. To avail the replicated dataset on the other server, you need to mount the latest snapshot and set the permission. But that's thats trade off. However, to the best of my knowledge, you can set a true HA on Nas4Free using carp. [ Very similar to TrueNas but UI is confusing ] . Corrections, if any, in above explanation , are most welcome.

[@densonngumo7028]

Have you deployed any of that in production? Trying to comeup with a HA storage solution for my VMs. I've used NFS with DRBD but testing HA failover wasn't seamless

[@LampJustin]

@@densonngumo7028 yeah as said in a different comment, NFS Ganesha + Gluster provide the best failover experience you can get on NFS

[@LampJustin]

Scale will come with support for Gluster ;)

[@Darkk6969]

Synology offers brain dead way of enabling HA between two Synology units whereas TrueNAS there are alot of steps to go through. Now if they come up with a HA wizard then it'd be a snap.

[@mithubopensourcelab482]

@@densonngumo7028 Off course... DRBD is not necessary. Simple NFS is more than sufficient.

[@UltralifeTech]

Is there any way to set this up with multiple sites over a VPN? So if one site loses power/connection, the other site will remain connected?

[@pepeshopping]

If you can get a Gigabit or faster link, sure.

[@UltralifeTech]

@@pepeshopping what about just for SMB shares? They don’t NEED fast transfers

[@jasonh2904]

Does anyone know how or if you can install the agent dvr program on truenas core? I need it installed on there for my security cameras but I don't know how to install it or if it even can be done. Could someone help me out please?

[@xaponetta]

You forgot to setup the Quorum server.

[@pepeshopping]

Not every HA has a Quorum server!

[@xaponetta]

@@pepeshopping You can easily set a switch or a router or any other device that is live 24/7 as a Quorum, split brain is an annoying problem when it occurs

[@attilavidacs24]

First