Search results

Are you agreeing or countering the argument? I'm not even sure.

You can have your zones locally and the auth servers externally. Keyword "hidden primary" I prefer to run my own servers so I have query logs.

No, why? Lanes have nothing to do with drives as I said. PCIe lanes are a really low level concept and totally transparent to the use case. You wouldn't get the idea that a graphics card in an x16 slot would only use one lane because only 1 monitor is connected. Such an idea is equally silly.

Drives don't utilize lanes. The controller does.

User experience has to suffer sometimes for the cash to flow.

PCIe 2.0 x4 is 2GB/s.

Use bs= of at least 64k. Why only 16k?

Could you guys at least read the papers? What would be the point if the client sends unencrypted data to the first node? That would compromise who the client is and what target it's trying to reach all in a single step. Also, it would be stupid. Roughly, it works like this: To send data...

It's not encrypted at each point. It's encrypted by the client only.

Do you know how a Tor hidden service operates?

The sound of naive world views being shattered.

Basically, all data you store is critical in one way or another. Otherwise, why store it? Personally, I don't understand how people can not care about corrupted files in the long run. Watching a movie and encountering a corruption that can spread over several frames visibly doesn't bother you...

Violating the license would be your only bet.

Don't tell me, tell Stevens. I just quoted him.

How about we go to TCP/IP Illustrated Vol. 1: "Packet" is just a general term for.. well.. packetized data. If you want to be 100% correct you say frame and datagram for L2 and L3, respectively.

Speaking of nitpicking, what do you mean when you say data goes "through" the 3rd layer? L2 data includes the L3 data so how can it go "through" L3? Or "within, between, and across layer 3" for that matter? Data doesn't go "through" layers as if the layers are somehow planes that the data passes...

I gave up trying to cram everything into a small space. What for? Just buy that spacey 24-bay rack case, load it with a dual-socket E-ATX board and never have to upgrade anything again except for stuffing in more HDDs. My dual-S2011 board draws less than 40W idle with one CPU and 32GB RAM and...

Dual-channel vs. quad-channel RAM. More cache. 16 vs. 40 PCIe lanes. Depends on what you need.

Duh, dunno what I was thinking.

You certainly want to filter spoofed source addresses.

A 10-drive RAIDZ2 has 8 data spindles, so 125MB/s per.

Uhh, just because all you know is VLANs doesn't mean they're the right tool for this job. There's absolutely no need for one VLAN per client. If you want the clients to be able to talk directly to each other, use simple port security so they can't for example assume the MAC address of your...

A 64-bit CPU is not twice as fast when doing the same task as a 32-bit CPU. Performing 64-/128-bit calculations on a 32-bit CPU is not half or quarter the speed of a 64-bit or imaginary 128-bit CPU - certainly not enough to go from 40MB/s to Gigabit throughput. Misunderstandig bit sizes this...

LOL! Nice troll.

No, if you're already paying premium for server-class hardware, it's royally stupid to not buy ECC for a peanut premium.

Hi, it seems this wonderful forum - between hour-long downtimes - doesn't save the visited status of threads now, which means I get the same posts as "new" even when I've read them. kthx

Yes, DNS is so simple, it takes a genius to understand. :)

What's the mechanism that Windows itself uses if you check "Update DNS" in network properties? Edit: Nevermind, brainfart. Cygwin+nsupdate and some tray tool? No idea. This belongs on the router not a client.

Get a DNS provider with hidden primary support and run your own local auth DNS that you can update using regular RFC2136 updates.

One of the main advantages of choosing a server platform is the ability to use ECC. I wouldn't ever recommend using non-ECC on a server platform. It makes no sense.

Well, uh, depends on what platform you want? X79 is consumer Socket2011, i7, no ECC C22x is Server Socket1150, Xeon E3 v2 (Edit: v3, not v2) C60x is Server Socket2011, Xeon E5 and E5 v2 You don't choose your chipset, you choose a platform depending on how much RAM and cores you need. I'd...

Why should SnapRAID have any problems with ZFS? SnapRAID doesn't know about disk blocks. It's file-based.

What does the ESXi console actually show? PSOD?

Yes, because ICMP and UDP are stateless. Currently, they go like this: Request: 10.2.1/24 -> 10.1.1.15 -> 10.1.1.10. This works. Reply: 10.1.1.10 -> 10.1.1.1 (because of 10.1.1.10's default route) -> 10.2.1/24 via 10.1.1.15 (because 10.1.1.1 has a route for that) They take completely...

OK, apparently I'm retarded. I was partly misunderstanding the problem from just seeing the PFsense interface. So, you have 10.1/16 on the PFsense box. 10.1.1.15 is your VPN endpoint. Why does traffic from the VPN to 10.1.1.10 even hit the PFsense box? It should go from 10.1.1.15 directly to...

Wait a second. You're doing VPN to 10.1.1.15? Does that box do NAT for the VPN network? Otherwise, your PFsense needs a route to the VPN network via 10.1.1.15. Edit: Nevermind, you got this route. Still looking. When you try to connect to 10.1.1.10, what does pfctl -ss | grep 10.1.1.10...

You're right. You can deduce the direction by guessing that 80 is most probably the destination port, but only if you know what interface 10.1.1.10 is living on. If you do tcpdump pflog0, you get the direction spelled out so there's no room for misunderstanding. The TCP:SA tells you this is...

Unless you have a state-policy if-bound, all rules should create floating state. I'm just not sure if that works as intended with the reply-to option. These firewall logs are also conveniently missing the direction the packet was blocked on. I would almost recommend dumping any fancy...

Check zpool get failmode <pool>

First hit for "TCP-in-TCP": http://sites.inka.de/~W1011/devel/tcp-tcp.html Unbookmark every site that says otherwise, now.