Hello! I’m trying to ping some lemmy instances to understand which one is the faster, so I’m just using the ping command:
$ ping lemmy.ml
PING lemmy.ml (54.36.178.108) 56(84) bytes of data.
64 bytes from lemmy.ml (54.36.178.108): icmp_seq=1 ttl=49 time=24.4 ms
ping lemmy.world
PING lemmy.world (135.181.143.230) 56(84) bytes of data.
64 bytes from static.230.143.181.135.clients.your-server.de (135.181.143.230): icmp_seq=1 ttl=52 time=58.2 ms
but if I try with certain instances:
ping vlemmy.net
PING vlemmy.net (109.78.160.70) 56(84) bytes of data.
it just hangs there, forever. if I try to ctrl+C it, it displays
^C
--- vlemmy.net ping statistics ---
13 packets transmitted, 0 received, 100% packet loss, time 12267ms
why does this happens? I can perfectly visit vlemmy.net from my browser so I really can’t understand whay is this happening
One possibility is that the server is configured not to respond to pings
This is most likely the issue. Maybe just measure HTTP response times instead?
what advantages can derive from this?
ICMP is one vector of a distributed denial of service attack. Also, even if not denying the service, it puts load on the network interfaces / routers of that server - everything behind the first firewall / router layer that would otherwise block the ICMP requests.
okay I get it, thanks!
Prevents some types of port scanning normally. Don’t know about other advantages
honestly, not much… other than warm and fuzzies for the person doing the blocking. ICMP sweeps are quick and cheap, so it stops a host from being found easily, but there are other ways to find abhost that are just as easy and cheap.
I think we block everything that’s unecessary, to minimise any attack surface area
Adding to the other answers, there isa bunch of server software that comes preconfigured to ignore ICMP.
ping nowadays is overrated anyway. If a server responds to ICMP and how fast it does it does not really say much about “how fast” a website is. It only tells you that a) ICMP requests and responses are not blocked and b) how fast ICMP requests get answered.
That’s it. It may not even tell you that a website is online because a load balancer may be responding to the ICMP request while the hosts behind it are offline.
People value ping responses way too highly.
httping may be a better tool to measure “how fast” a website is responding.
Also every major browser has a tool for timing and seing how long a site and it’s components load. You could test it with that but even then; load times will vary slightly depending on what the instances have to load.
But probably a better way than pings ¯_(ツ)_/¯
Correct, especially with all the dynamic loading and rendering websites nowadays do measuring in a web browser is waaay better than doing ICMP/ping requests or even httping requests.
It depends on what you are trying to measure ofc but ICMP/ping does not tell you almost anything about how fast a website is.
People value ping responses way too highly.
I beg to differ. Not everyone’s use of the internet is limited to http(s) - ping is an invaluable tool to determine round trip times of the underlying network infrastructure & therefore assess e.g. the potential throughput of TCP based protocols for given window sizes. Also, to assess delay in UDP based communication.
Yeah that’s what I was saying though. OP was not measuring that, that’s my point.
Ping is still very good because IMCP it is built into almost all network interfaces. So it it is a great tool if you want to check if a host is reachable from the device. But yeah, it is different than anything that has to do with websites.
httping may be a better tool to measure “how fast” a website is responding.
this works well, thank you! the results are similar to the ones obtained with ping, but vlemmy.net and other instances that ping couldn’t reach are correctly measured. seems that lemm.ee is the fastest for me
ICMP echo requests/responses may be blocked - usually by a remote endpoint firewall.
The server’s firewall could be blocking them?
If it’s a VPS, most providers’ firewalls will block everything by default, and you have to specifically choose what ports to open or transmission protocols to allow.
I personally have all my VPS’s set up to drop pings from any unauthorised IP addresses
As others have commented, blocking is the most likely reason. That said, responding to ping requests typically is deprioritized when the networked device is loaded heavily and requests get dropped.
Don’t rely on ping to tell you that. That only tells you how quickly your packet reaches that particular server, which could change based on your ISP, your location, their ISP, their location, and — as you saw — their server’s firewalls.
Plus, ping only tells you how quickly the hardware responds. The site could have slow web server software, be slow to serve web pages due to user load, or have a bloated web site full of megabytes of JavaScript (though I hope that’s not an issue in the Lemmyverse!).
Maybe the server’s firewall drops ICMP packages?
There’s a firewall blocking ICMP echo-reply requests on the other end. It’s totally normal for servers to block ping requests.
Ping just measures the response time between you and the server. The Lemmy application doesn’t even know about pings. It’s the computer network card responding to them, unless the computer is configured to simply not respond to them.
You can test the response speed of the web server using curl, but it really doesn’t say much about performance either. It’s difficult to do good performance tests on stuff like this.
I think the closest you can get to a good performance test is to use a virtual browser like https://www.selenium.dev/ and simulate browsing to a new post, clicking some link, and so on.
deleted by creator
