Not sure this still works these days, but this is what I used in the past: GRC | InSpectre
Hmmm, I’ll have to give it a once over.
My Employer’s payment portal only works through Internet Explorer (for the record I am a college student stuck at a retail job).
Does anyone have any experience with smtp2go for email alerts? We are currently using a Gmail account to send alerts for out UPSes, Printers etc. If you have recommendations of other email alert software I would be welcome to that as well.
Thanks
Zen
Pro tip for anyone who might run into this problem down the road. Hairpin-NAT requires a firewall rule, but not the one you are thinking. It took me ages to figure this one out.
Say you had a web server hosted at 1.2.3.4 and needed to port forward to an internal address 192.168.100.100. Easy enough.
/ip firewall nat
add action=dst-nat chain=dstnat dst-address=1.2.3.4 dst-port=80,443 protocol=tcp to-addresses=192.168.100.100
As the packet flows through the router this will rewrite the destination address from 1.2.3.4 to 192.168.100.100/24 and send the packet on its way. Since the “from” field of the packet is still intact, the response from the server to the client is unimpeded.
This presents a problem though. All public traffic will work fine, but if you try to visit 1.2.3.4 in your web browser you will fail to connect. This is because the connection to 1.2.3.4 is rewritten on your router to 192.168.100.100 and that server responds to the “from” address in the packet directly to your local LAN IP. Since the traffic returning to you is not from an established connection, the data is dropped and it looks like the server is not responding. How fix?
I spent all morning banging my head against the wall trying to figure out how to properly masquerade traffic destined to public addresses bound to my router’s public interfaces. It turns out I was barking up the wrong tree. The actual rule to add is:
/ip firewall nat
add action=masquerade chain=srcnat src-address=192.168.100.0/24 dst-address=192.168.100.0/24
(this could be shaped up a bit to make sure the traffic is actually local)
Add that as the first rule and the one above as the second and it will work. Why?
When the packet hits your firewall it matches the first rule above and the destination address is re-written. If a change has been made to a packet it gets re-sent through the router firewall rules. At least, that’s what the Mikrotik packet flow diagram says.
Next, since its source and destination address are now both local, the other firewall rule picks it up and applies the masquerading rule to it. Now the packet appears to the server as coming from the router itself, and the router will un-masquerade the response and send it to the original client.
Taa daa, hairpin-NAT.
Seems legit as it follows the order of precedence.
So I have this Arista 7150S switch and I ordered the rails for it. The rails don’t come with the screws to attach it to the switch… Based on photos, this is the case for all the rail kits (mine was new in box). Any ideas on how to source these screws? I know there are ways to figure out the size/thread/head-type but I’ve never had to do it.
Nvm, I got it. They’re m4 and pan head will work although not sure that’s what came with it originally. The depth is flexible which is what I was mainly worried about. Ordered some m4 5mm should be here tomorrow.
Should be fine. So I am spoiled because all USG orders come with all that stuff. And we have extras of everything just in case we get sent the wrong stuff.
Does anyone know what the -0in the following piece of code is doing. It is really hard to specify this into a online search request.
while some_function -0; do
sleep 0.5
done
It’s just passing “-0” to some_function as $1.
I decided to get fancy and automate more stuff by salt-ifying it.
First up, automating certbot renewal for haproxy since its a PITA and is a manual step to create the combined file at the end.
I’m working on similar a thing
I might still have my certbotbot if you’d like me to dig through some of my old files.
Dunno if it’ll help tho, it’s all bash and ruby.
Nice!
Let’s compare results when we’re finished.
Does it also hook into cloudflare?
PS: apparently they drank the Cananical punch and are only supporting new versions of certbot via snap.
So I spent the better part of my afternoon wading through that mess.
It felt sinful installing snapd on a redhat system.
Oh no. Time to look into acme.sh next time I work on my self hosted items.
No, we didn’t use cloudflare where I was using my certbotbot.
Thinking of upgrading my trashcan home server with an HP i5-3330 trashcan from a non-profit resale store. No HDD, M&K, or monitor.
Is there any other way to know if it works beyond a posting and hoping the lights come on ok?
Lots of things you could do, depending on what you have.
There are small, portable monitors and mini keyboards the size of 7" tablets or so that could do the job. But used/thrift type stores usually have have a monitor keyboards and mice around that can be borrowed for a bit to try out other available kit.
If it’s really a server system, it might even have serial console redirect enable. I’ve got my old Psion 5MX with serial port and terminal software that fits in my pocket, so I’d take that with me. Then a bootable USB drive that goes right into SystemRescueCd with serial port consoles enabled.
You could take your laptop and a network cable in with you. Bootable SystemRescueCD USB drive again, but configured to bring up networking with a static IP address and SSH enabled. Configure your laptop on the same network and SSH over into the HP.
Wait, why?
I have set up (but not yet automated) certbot wildcard cert via authoritative dns server with dnssec. The only config needed with the registar is name servers/glue records and ds records. I’m still early in the process of figuring it all out and testing but I believe the only recurring task with the registrar (barring IP changes) is to update ds records but only upon ksk rotation which is usually infrequent (year+).