Hot answers tagged

18

Reading the papers on HFSC and its cousins is not a good way of understanding it. The primary goal of the HFSC paper is to provide a rigorous mathematical proof of its claims, not explaining how it works. Indeed you can't understand how it works from the HFSC paper alone, you have to read the papers it references as well. If you have some problem with the ...


8

netstat -an | grep ESTABLISHED | wc -l This will count all opened sockets in the system and will output just the total. You can of course also change ESTABLISHED for whatever you need, for example a port or a communication status like CONNECTED or LISTENING.


5

Answer: DRR does not drop packets itself. To get the desired behavior, add child qdiscs to the DRR child classes like pfifo limit 50 to cause the child qdisc to drop packets instead of queueing them effectively indefinitely. The solution was found here: linux.org.ru thread Reference: man tc-drr NOTES This implementation does not drop packets from the ...


5

This is impossible to speak authoritatively on, but in my experience, the vast majority of ISPs ignore QoS and DSCP headers, and with good reason. If they respected those headers, they would quickly become meaningless as everyone just starts to mark all their packets as EF or AF or somesuch. The only way to have true end-to-end QoS is to run the entire ...


4

To watch live MySQL opened connections through UNIX socket, run (as root): watch -n1 'netstat -np | grep -i mysqld'


4

As long as the companies only share the internet connection and have no need to share other resources (like file servers), I clearly would favor option 1. If you have separate companies and have become the administrator by chance rather than an external, binding decision for every of the companies, one of the things you would want most would be a clean, ...


4

The IP range 78.31.8.0/22 also belongs to spotify.com so the complete list would be 78.31.8.0/22 193.182.8.0/21 193.235.232.0/24


4

It depends on the type of QoS you want to use. The world of QoS is vast and complex and there are many different techniques that can be combined in different ways to give a "guarantee" of service in each situation. Actually I'd say that reserving bandwidth for VoIP is a type of QoS policy. If you ask for a type of QoS policy in particular, you should ...


4

Problem is with high contention, packets begin to queue up and VOIP packets cannot sit and wait. Are you sure you configured destination ports appropriately for traffic exiting your site? You can't packet shape ingress, only egress. Maybe you configured those destination ports for your LAN side which might include a lot of non-VOIP traffic as well ...


4

The filter priority isn't family specific. So, your commands to create filters should look like: /sbin/tc filter add dev eth0 parent 1:0 prio 1 protocol ip handle 1/0xF fw flowid 1:1 /sbin/tc filter add dev eth0 parent 1:0 prio 2 protocol ip handle 2/0xF fw flowid 1:2 /sbin/tc filter add dev eth0 parent 1:0 prio 3 protocol ip handle 3/0xF fw flowid 1:3 /...


3

drr is a scheduler, you still have to allocate different bandwidth to classes with htb. I guess you though specifying quantum as 600 and 1400 will give close to 1:2 ratio. In fact it will not. You may get close to that ratio on your config only in case of congestion, e.g. create multiple UDP streams, and then measure two in question, but it's still not the ...


3

I would favor Option 3 personally: Install a proper "Big Managed Switch" like you have in Option 2 Put each company in their own VLAN. For the uplinks you have a few choices: Give each company their own uplink (modem, etc.) inside their vLAN like you have in Option 1. Install a decent firewall like PFSense with an interface in each vLAN and funnel all ...


3

Typically the application doesn't control the bandwidth allocated to specific sockets, nor does it know how much the kernel is sending its traffic versus that of other applications. The utility tc (part of iproute2), combined with iptables setting marks on packets and connections, can be used to implement an arbitrarily complex hierarchical token bucket ...


3

Option 2 - will give you the opportunity to buy a L3 switch with more capability, capacity, flexibility and resilience than option 1. Overall performance will is likely to be higher, you'll have the option to use PoE ports for the phones if you wish and you'll have the option to run some form of network management tool tool so you know what's going on and ...


3

The question is rather old, but just in case someone comes across a similar problem. Parameter all can be used for protocol instead of ip to also filter for other traffic like arp. tc filter add dev $IF_LAN parent 1: protocol all prio 7 u32 match u32 0 0 flowid 1:190 Also match u32 0 0 is a shorter expression to filter all traffic.


3

You don't need to do NF_DROP verdict on incoming packet and resend it via raw socket. You can hold packet in your own queue, and verdict it later (say, after some delay). I wrote a simple traffic shaper in Linux userspace with this technique: https://github.com/vmxdev/damper/ Using only NFQUEUE for delaying packets simplifies iptables rules. You don't have ...


3

I think you are misunderstanding how QoS works. Lets say you have two applications running on a server (app A and app B) and either one is maxing the bandwidth. You decide you want to make sure App A gets more of the bandwidth. In this case you would define two polices (lets say using exe name), you give App A a DSCP value of 10 and App B 5. You assign a ...


3

From @akostadinov's answer: "One can create multiple SSIDs in the router. Then assign a different QoS value for each SSID." The SSID here means that you'll have the QoS set based on the SSID priority value. https://fccid.io/Q78-ZXW3512C/User-Manual/User-Manual-1928235 And according to the manual, it's simply that higher number will have higher priority, ...


2

I'm not sure NFS can be "hardened" against what amounts to a DDOS from a cluster. If you really need that, using something else to access persistent storage will be easier. Given your setup, I would suggest doing the "QoS" on the cluster engine level. Configure a resource "io_heavy" limited to say "10" and make your users request "1" of this for IO-heavy ...


2

If your computers are on a Windows domain , you can use group policy QoS settings to assign DSCP values to traffic generated by the Skype executable file (skype.exe)(actually, you can do the same with local policy on a workgroup computer, you just might need a registry key added too). DSCP 46 I believe is what's usually used for expedited forwarding. Then ...


2

It is possible. I believe tc(8) man is not comprehensive. Citing from man7.org/tc(8): tc filters If tc filters are attached to a class, they are consulted first for relevant instructions. Filters can match on all fields of a packet header, as well as on the firewall mark applied by ipchains or iptables. Just checked on my Linux 5.0.3-zen1-2-zen ...


2

That's a super long question! The bible for this sort of stuff are the SRNDs (Solution Network Reference Design) produced by Cisco - in your case http://www.cisco.com/en/US/docs/solutions/Enterprise/WAN_and_MAN/QoS_SRND/QoS-SRND-Book.html I hate to say "RTM" but in this case the topic is so complex and vast that we'd need more specific questions to fit an ...


2

I guess you mean 8-bit TOS field in IPv4 packet header. Have you alread read LARTC HOWTO? This is the guide you definitely need to read. Briefly, you need to 1) Define QDISCs and CLASSes in which traffic is classified, prioritized and shaped (egress traffic can be shaped only !!!). In general, any packet which needs to be sent is enqueued to QDISC of ...


2

Do you trust your workstations to not submit data with a fake QoS header just to get more speed? You can not avoid that but on a VLAN you can prioritize the VLAN. Also, on windows doing QoS on the workstations can be tricky ;) That said, I would say in most cases you do not even need QOS; 30 is a small number of phones and workstations.


2

I recommend using a separate VLAN whether you use QOS or not. A separate VLAN allows you to set different DHCP options (such as Boot Server) for your phones without affecting your workstations.


2

It's all because of net.inet.ip.dummynet.io_fast=1. This variable turned on fast packet processing: while bandwidth is not exhausted, all packets are forwarded directly through the bridge. Shaper is bypassed completely - and no packets counted in the DUMMYNET. That behaviour is intended specially for heavy loaded gateways for CPU offload for the cost of ...


2

Windows Server 2003 R2 has already hit mainstream support end-of-life a few years ago, and is scheduled for complete end-of-life in a few months. I would not use it for anything, much less a router replacement. You're going to be much better off with an actual router or firewall. To answer your odd question: Yes, with the right configuration and software, ...


2

Regarding your question on the policing/limiting : The input and output are being policed to 1000000 bits per second with a maximum burst of 1500 bytes. The burst value is the max number of bytes in a sustained burst that is allowed before traffic throttled to the specified rate of 100000 bits per second. This class belongs to a policy-map which can be ...


2

What is the specific model of the router provided from your internet provider? Do you get a few static ip's with your internet package or just dynamic? I am using a CISCO ASA series to do a very similar configuration. [We divide bandwidth based on branch offices] Your HP device is a switch it would be better to setup two routers behind the providers ...


Only top voted, non community-wiki answers of a minimum length are eligible