I run a NAT router on Debian 10 which takes only a few seconds to reboot, way less then 120 seconds, more like 3. I'd like to keep established connections so that a reboot of the router does not break natted TCP connections, but I don't have the capacities to run two synching routers. How do I persist the conntrack tables on a single machine?
When configuring the management IP address of managed Cisco switches, I have to specify the broadcast address of the management network next to their address AND their netmask. Of course this is always the last address of the subnet.
Why do I have to specify it? Why doesn't it calculate the address on its own? Are there situations or reasons to use a different one?
Imagine you are browsing to a HTTP URL that does not exists. You'll get a 404 which often contains an explanatory body.
Now imagine you are calling someone using VoIP with SIP and use a non-existent addressee. Is it possible to return an error AND to establish an RTP stream with an explanatory audio message?
I have setup a Postfix as a relay for an embedded device with a severely broken STMP implementation. Therefore I need to make Postfix to have AUTH enabled and to accept ANY combination of username and password. As of now I have it relaying my mail to another relayhost for the whole subnet, but when the device tries to AUTH PLAIN, Postfix answers with 503 5.5.1 Error: authentication not enabled which make the embedded system interrupt the transmission.
How do I make Postfix accept ANY user:password combination?
I found various lists of MTUs for different links, but they never contain UMTS, LTE, EDGE, HSPA, GPRS.
What are the MTUs of the above mentioned technologies?
I'd like to setup dnsmasq as DNS-only server for blocking certain domains. But instead of having it return 127.0.0.1 or 0.0.0.0 or such, I'd like it to respond with a No such name. just as if the name does not exist at all.
How to I filter certain domains?
I'm using a dual stack connection at home and have a DS213+. As my public IP (both v4 and v6) is dynamic, I'd like to only specify a suffix in my Diskstation's network configuration. My router supports DHCPv6, offers IA_PD and IA_NA.
My ISP delegates a random /64 to my router each time I connect and I want my Diskstation to use this prefix plus a static suffix. I did NOT find any option to do this.
My setup is the following:
Synology Disk Station as iSCSI target
OS X 10.8 as client
Debian as client
I don't like to know that I should use WebDAV or similar file sharing protocols.
I'd like to access the same target from at least the both above named devices. What file system should I use? I guess it should be a cluster file system, but which one should I use for best compatibility?
I'm on a non-manageable switched 1 GbE network with mixed 100 MbE and 1 GbE hosts.
Is it possible to tell if a certain host has a certain link speed without looking at the status LEDs of the switch and without access to the hosts?
This is a follow up on Why Doesn't Ping Show RTT?
I had a look at BSD, Linux and BusyBox ping implementations. All three share the feature that they calculate the round trip time by sending a timestamp as echo request, receive it back as echo reply and then subtract the received timestamp from the current time.
Obviously the target host could modify the timestamp before it echoes it back, giving weird ping output. In addition to that, icmp protocol specification does not talk about timestamps at all. It's just stored in the payload field. If you modify the size of the payload to be too small to store the timestamp, ping is unable to calculate the round trip time.
I already patched a ping implementation to save the timestamps and not to rely on the echo. It seems to work perfectly, but I'm unsure if I'm breaking anything (in real life scenarios, the patch totally obeys the standards).
My questions:
1) Do I break anything?
2) What reason could the designer of ping have had for implementing it in such a non-obvious way?
I'm on OS X 10.7 and just found out, the
ping -s 10 host
does not work as expected. Its output does not contain the round trip time:
PING host (IP): 10 data bytes
18 bytes from IP: icmp_seq=0 ttl=248
18 bytes from IP: icmp_seq=1 ttl=248
^C
--- host ping statistics ---
2 packets transmitted, 2 packets received, 0.0% packet loss
It works if -s's value is 16 and above. I also experienced this on Linux, but with a smaller value. I read man ping, but it does not mention a restriction.
I'd like to change the MTU per host as I have a mixed infrastructure where not all devices support the same MTU. I'm on OS X 10.7 and like to set some MTU per destination host and a default one.
How do I make Postfix to reject all mails if the client is not authenticated and to sender is @mydomain? Is the an easy way to do this?
Max