Should I Upgrade to v16.6.2 - Home User (HomPro Subcription)

[BryanC1968]

I got up a little after 5:00 am this morning and did a fresh install of Untangle 16.6.2 before anyone else got up this morning so as not to disrupt anyone internet access... I restored my backup I saved before beginning the process and As most everyone has reported, my interfaces were all rearranged... Had to spend some time getting that all sorted out and even had to redo my Network Address Configurations I had configured on each interface in order to get everything lined back out the way i had it before... But everything seems to be work as far as I can tell for now... Only time will tell if I run into any issues...

[BryanC1968]

After 13 hours of uptime on v16.6.2 I have noticed that the system appears to use less memory than v16.5.2 did... My system has 16GB of ram and v16.5.2 seemed to hover around 2GB of ram usage, but now v16.6.2 is currently hovering around 1GB of ram usage... Running all the same apps and same config... Actually I'm testing the 30 day trial of Wireguard VPN and Threat Prevention that I did not have installed before... I noticed after upgrading that I was able to install and try these apps again even though I had previously tried them with a 14 day trial before... I have been considering upgrading my subscription to the new Home Protect Plus... Currently on an old Grandfathered HomePro Subscription...

[soldier]

I'm currently on 16.5.2. I did a configuration backup and started to upgrade. The following message appeared.



This is the first time I'm getting this message and I've been using Untangle for 3+ years.

[gravenscroft]

This is the first time I'm getting this message and I've been using Untangle for 3+ years.

We're aware of it: Issues with downloading updates

[BryanC1968]

I'm currently on 16.5.2. I did a configuration backup and started to upgrade. The following message appeared.



This is the first time I'm getting this message and I've been using Untangle for 3+ years.

I downloaded the ISO image and did a fresh install, and then restored my backup from 16.5.2... Instead of doing an inplace upgrade...

[sky-knight]

I downloaded the ISO image and did a fresh install, and then restored my backup from 16.5.2... Instead of doing an inplace upgrade...

Honestly, this is almost my preferred method to handle OS level upgrades, they can go so wrong! But I am notoriously paranoid, in reality they usually work.

[soldier]

I've managed to download the files and upgrade. So far so good.

[sky-knight]

You do not have a choice on the upgrades, and you cannot skip releases. NGFW must be upgraded incrementally. What you can do is wait a few months to see if things settle down a bit.

I have my fleet on v16.5 still, but that's largely because I haven't had the availability to be present during the upgrade process for the devices in question. v16.5 to v16.6 is an OS level upgrade, so I will not push that button unless I have a configuration backup in my hands as well as installation media at the ready.

As I type this I'm finally getting around to updating the appliance I use here at my office, I'm not expecting much of an issue but I will be loaded for bear just in case.

*Edit* Or I won't... because:

Since I have no admin alerts about connection issues, and here I am posting about it, I'm left to assume that Arista disabled the NGFW upgrade servers for some reason again.

*Edit 2* Upgrades have been disabled for awhile now, nothing to do but wait until they get the servers sorted.

https://forums.edge.arista.com/forum/general-forums/announcements/396192-issues-with-downloading-updates

[jbhur]

The full details are all here: https://wiki.debian.org/NetworkInterfaceNames

Anyone that lives on a Debian based system deals with this reality. You either make the .link files to anchor your NICs or you risk the flags shifting on each and every reboot. Even a stable machine with a sane BIOS can get enumerated differently when the kernel changes.

Thanks for this.
I have grief rebooting my V16.6.2 installation ever since upgrading many upgrades ago. After a reboot of Untangle, the network device names do not match the devices' actual MAC addresses (as reported in the Config > Interfaces > Remap Interfaces) and chaos ensues. Network device names and MAC addresses always align only if Untangle is cold restarted - a soft reboot now never works.
I had always attributed this behaviour, in part, to the unconventional installation I'm running. Untangle runs in a Hyper-V VM. Windows' frequent updates, once installed, often require Windows to be restarted. If I direct Windows to "update and restart", Untangle returns with its network device mappings screwed up. However, selecting "update and shutdown" always results in correct mappings once the bare-metal host is manually restarted. This annoyance is mitigated by delaying Windows restarts for a month or so - not a big deal.
Maybe the link you provided will fix this if I'm ever able to understand the instructions.

[RainCaster]

My CPU Load went up considerably after the install. Why?

[reachmedia]

I think for this new build .... better to setup as a new instance and import the config settings over. Perhaps Arista can do an old backup file (16.5.2) conversion to import for 16.6.2

[sky-knight]

Thanks for this.
I have grief rebooting my V16.6.2 installation ever since upgrading many upgrades ago. After a reboot of Untangle, the network device names do not match the devices' actual MAC addresses (as reported in the Config > Interfaces > Remap Interfaces) and chaos ensues. Network device names and MAC addresses always align only if Untangle is cold restarted - a soft reboot now never works.
I had always attributed this behaviour, in part, to the unconventional installation I'm running. Untangle runs in a Hyper-V VM. Windows' frequent updates, once installed, often require Windows to be restarted. If I direct Windows to "update and restart", Untangle returns with its network device mappings screwed up. However, selecting "update and shutdown" always results in correct mappings once the bare-metal host is manually restarted. This annoyance is mitigated by delaying Windows restarts for a month or so - not a big deal.
Maybe the link you provided will fix this if I'm ever able to understand the instructions.

Virtual NGFW installations must have static MAC addresses, some hypervisors will make them dynamic... you must ensure that's off.

After that, it's a matter of making files in /etc/systemd/network based on this template

#/etc/systemd/network/10-persistent-net.link
[Match]
MACAddress=01:23:45:67:89:ab

[Link]
Name=lan0​

So for example I'll make a file named 10-persistent-eth0.link and it'll look like this

[Match]
MACAddress=01:23:45:67:89:ab

[Link]
Name=eth0

And on the next reboot udev will hard assign the NIC with that MAC address to eth0.

Then there's 20-persistent-eth1.link, and 30-persistent-eth2.link, and so forth. One file for each NIC, once those files are written, Debian will always boot with the NICs in the predicted place and the UVM can work its magic. All the details are here: https://wiki.debian.org/NetworkInter...NG_.LINK_FILES

And yes, everyone needs to do this before you upgrade to 16.6 because if you don't... you WILL have interface flying around. And when that happens you'll be on the physical console in a mad panic trying to figure out where External went, and once you have that fixed you'll realize your VLAN virtual interfaces have the wrong parents, and IP address aliases don't work until they're pulled and put back... it's a huge pile of juggling in config -> networking -> interfaces to get the platform back online.

The good news is once you've done all this on the new kernel it'll not be an issue until the next one. But the only way I know to future proof any installation, is to make those files. Even that though is not a guarantee because Debian has flat lost its mind in this regard.

[RainCaster]

I'm seeing a similar set of numbers on my system. Here's the plot for the last 7 days, two of which are on new version. This was taken with the latest version of Zabbix. The machine is a HP ProLiant DL360-G7 with 24 cores, 64GB. Yes, I have room to spare, but I'm still concerned.

EDIT:
Interestingly enough, it looks liike Debian is well aware of this, as my fans all went up by 10% with this update. Also, CPU core temps all dropped 5C with that fan increase.

[BerkeleyIV]

I'm seeing a similar set of numbers on my system. Here's the plot for the last 7 days, two of which are on new version. This was taken with the latest version of Zabbix. The machine is a HP ProLiant DL360-G7 with 24 cores, 64GB. Yes, I have room to spare, but I'm still concerned.

EDIT:
Interestingly enough, it looks liike Debian is well aware of this, as my fans all went up by 10% with this update. Also, CPU core temps all dropped 5C with that fan increase.

Thanks RainCaster. Appreciate the opportunity to see confirmation data from other systems.

[gravenscroft]

I think for this new build .... better to setup as a new instance and import the config settings over. Perhaps Arista can do an old backup file (16.5.2) conversion to import for 16.6.2

A backup file is always valid for the version it was created on and one minor version newer, so a 16.5 backup file works on 16.6.