check dmesg for undervoltage complaints?
If you’re not getting enough power, you’ll get undervoltage complaints.
If you aren’t getting any, you’ve got a bad unit and it should just be RMA’d. (or replaced)
1 Like
Never have that problem with Cisco /s
New box or older? If older assuming no changes recently
1 Like
Successfully upgraded from 29->30
3 Likes
T.T how much is a yearly RHEL subscription?
Oof, $350USD/yr for self-support.
Sign up for a developer acct… It’s free
2 Likes
(┛◉Д◉)┛彡┻━┻
They might as well go straight for 8.1 at this rate.
1 Like
“Most of our users on on the 7.x so we deferred 8”
No shit, because you haven’t released 8 yet.
Once they do its going to spread like wildfire.
1 Like
So I disabled jumbo frames (had used but no longer using), 802.1X and a “VLAN-Only” network and the switches appear to be working again without constant reboots.
Basically just cleaned up a bunch of crap from previous testing I had done at home. I was using the vlan to patch a port from my cable modem to my office so I could onboarding client networking eq with a temp public IP. This seems to have caused some other issues on the Unifi network, so I’m guessing it was the culprit here. It is a non-standard use case for Unifi.
Well I just got the word, new job. Going to be the sysadmin for a 150 user VC Firm. A one man team!
4 Likes
I haven’t moved yet. I probably will when I start this job do a fresh 30 install. Assuming they give me a new laptop or computer
Not so bad I guess…
2 Likes
Unless you have prior authorisation to do this from management, i would strongly suggest against this course of action.
And if you’re seriously wanting to test your monitoring, getting a pro to do some pen-testing might be an idea…
1 Like
I use /22s a little bit for this reason (on larger sites). Broadcast domains, etc. not so much of a problem with modern gig+ networking equipment, also not an issue with 100 meg. Also helps that most modern network protocols in use today aren’t totally fucking retarded.
Side-note(s) regarding IP subnetting… on my WAN i’m using /16s per site now.
A site route in my environment (irrespective of size) is
10.X.Y.Z/16 mostly.
X = site number
Y = VLAN number
Z = host
If i’m on a site that needs one or more /22s i block out several Ys for that particular VLAN across ALL my sites (for consistent network addressing from site to site). As above, even if i’m not sure, i’ll block out a /22 just in case. a /24 can get quite small if you like to say, reserve the bottom half or 100 hostIDs, etc. for statics or have more than a couple of hundred devices on a site…
I have a big excel sheet with a template that documents all this. A new site = allocate a new number for X, and that’s the IP scheme for site, for every VLAN/service we host. My home network also has a site reservation in this spreadsheet (10.48.x.y = home) so it can be joined to the corporate network and have working routing without any NAT bullshit if required at any point. As does my VM workstation hosted test environment (10.44.x.y = VMware Workstation lab). Keeps your routing table sane…
As the admin/architect of a 20+ year old international IPv4 network at this point (and i was involved from day 1 as a contractor in 1999 back when i knew jack shit about enterprise networking and this business had only 3 sites and no VLANs 
), i’d suggest to think about doing something similar from the get go in any private IPv4 network using 10.0.0.0/8 IP space.
Otherwise you’ll end up with a bit of a mess. Plan ahead for multiple sites (even if you think it will never happen) and multiple VLANs and aim for consistency across them. Makes automating shit easier, and remembering shit easier too. Even if you set up say, your parent’s house - set them up with a different siteID to yours, and you’ll have no NAT bullshit to deal with if you ever want to set up a VPN between them.
Had hopes we’d be running IPv6 at least in new networks about 6-7 years ago, but adoption amongst MPLS service providers is… slow. 
2 Likes
Pftt we aren’t talking about X299 here.
2 Likes
I also do this, exactly as described.
For my openvpn tunnels, I use 192.168.a.b addresses where a is local site and b is remote site, and I use a block of unused ports with a similar scheme. It was great how many thing became self-determined once I implemented that addressing formula.
1 Like
Yup. Makes things programmatically generated.
Also I might add - this scheme wasn’t implemented from day one (learn from my mistakes! 
) and I have a heap of non compliant legacy shit.
I’ve blocked out various ranges to prevent clashes with existing infrastructure, but moving forward that’s the way I’m going.
I reserve a few VLAN numbers for point to point links rather than using 192.168.x.y. I’d need to put them in a vlan if they need to go into a switch or vswitch anyway so…
1 Like
Do you use private vlans at all, where there’s no communication between the members (only gateway)? I’ve been meaning to set this up for my oobm vlan and maybe others. It throws a wrench into the vlan numbering though because you have to specify multiple.
Not yet but planning to for our workstations. On the end of a long list of shit to do…
1 Like
Cool. Let me know what you come up with for the vlan numbers if/when you do and I’ll do the same. I’m still unsure to what extent I can trunk pvlans across multiple switches on Ubiquiti Edgeswitches, so we’ll see if it’s even possible.