The journey that started back here when I seriously started looking at components for FreeNAS0 almost seems complete so I thought I’d blog about some finishing touches and my upgrade to 11.3.
I’ll start with the upgrade, which I guess I’d been putting off for a couple of months. 11.3 was released on 28 January 2020 and having just about finished my build and having everything running just as I wanted, I wasn’t feeling brave enough to jump on the first release.
At the time, I was thinking maybe the U2 release would be safe, although as it turns out I’m braver than I thought as 11.3-U1 was released (just less than a month later) on 25 February 2020 and I took the leap to upgrade on 9 March 2020.
I’d been keeping an eye on the forums, and while most reports seemed very positive, there were some of broken jails and VMs and some issues with replication which has been revamped in 11.3. I had a bad experience with broken replication in 9.3.1, which left me stuck there for almost 12 months, but in the end, I need not have worried. It does work pretty differently, and I struggled to get my head around it when setting up a new dataset to replicate between FreeNAS0 and 1, but I got there in the end. Not sure it’s any easier though!
I started the upgrade about 10:00 and pretty much left it alone for an hour or so while I tried to keep myself busy. It’s amazing how difficult that is when my FreeNAS box is down with no emby and no internet (without messing with DNS settings). Anyway, an hour or so later, I picked up my laptop to see what was happening. I had internet access, which meant that my Pi-Hole VM had restarted and was doing it’s DNS stuff. I had the Nextcloud icon in the Menu Bar, which meant at least one of my jails had started up. I had lots of e-mails about system alerts, but generally, these were duplicates saying things were down and had come back up.
I did a little more investigation from my iMac and was happy to find that all seven of the jails I was expecting to restart had, and were all working fine. The four VMs has all restarted, and while I had a few minor issues with the Docker VM, these were easily solved and not directly related to the upgrade. Replication also looked fine, although that wouldn’t really be tested until the next day when all my scheduled snapshots should have backed up to FreeNAS1.
It had gone so well that I thought I might as well update FreeNAS1 too! The WebUI for this didn’t come back up after ~45 minutes so I needed to dig into the IPMI console to see what was wrong. For some reason, the network address hadn’t been assigned, but I corrected that and FreeNAS1 came back up and again everything looked fine. While I was updating I also took the opportunity to replace a failing fan, which I’m kind of regretting now as the Noctua fan I’d bought to replace the Fractal Design one is way too loud. I might end up changing that, but If I do I’ll blog about that elsewhere.
Replication worked the next day and so far, 2 days later, almost everything appears to be OK. I appear to have an issue with my Crashplan VM shutting down, but I did update that, so don’t know if its related to the upgrade or not? I need to investigate but haven’t found the time yet and it’s easy enough to restart.
So what about the finishing touches? Well, to be honest, I still haven’t really got to the bottom of everything, but those things outstanding aren’t that important and will just need to wait until I have a little more time. I was also hoping the upgrade might help with one of two of them.
I’d blogged about switching Time Machine from here, but I never really followed up on it. My experience is that Time Machine over SMB is no better (or worse) than over AFP on FreeNAS. While it was possible to set things up on most machines, they don’t continue to run in the background all that well and even as I type I’m looking at a notification on my MacBook saying ‘No Backups for 44 Days’. Time Machine backups are a belt and braces approach, but I have other strategies in place to ensure my most important data is backed up regularly, and across multiple devices and locations. Time Machine not working doesn’t create any risk of data loss, just additional rebuild time should a device fail.
SMB shares seem to be a little more robust that AFP shares, at least in terms of maintaining their connection. I don’t have to restart the SMB service very often, at least not as often as I did for AFP.
I’m still having a few issues with Scheduled Tasks, or at least the SMART test layout, which is giving me some of the information, but not everything that I got on FreeNAS1. This is a minor inconvenience though, as I’m getting enough information to know if I’m having a problem, and to be honest after 3 months I’m not expecting anything significant yet!
I’ve configured Alerts to work with a Mattermost server which I blogged about here, and that’s working quite nicely, or is at least for FreeNAS. I’m also trying to send alerts from the emby jail, and they are not working, but trying to work out why is proving a challenge. Again, a nice to have thing so not the end of the world.
Lastly, I needed to tidy up the cabling inside my machine, and also reapply some of the thermal tape required to disable the 3rd pin issue.
Since I drafted this about 2-3 weeks ago, I have been having a couple of issues with the VMs, although nothing too bad. A couple of times my Crashplan VM has quit, but in doing so it shows all of the VMs as stopped in the FreeNAS WUI. The fact is they are still running, even though it’s showing they are not. I’ve logged this as a bug here, and will see what the iXsystems team come up with. No rush, as I simply restart the Crashplan VM, but in the interface, it says that’s the only one running, when the onlyoffice, docker/portainer and pi-hole ones are still up.