Max MQTT connections
Asked Answered
B

1

10

I have a need to create a server farm that can handle 5+ million connections, 5+ million topics (one per client), process 300k messages/sec.

I tried to see what various message brokers were capable so I am currently using two RHEL EC2 instances (r3.4xlarge) to make lots of available resources. So you do not need to look it up, it has 16vCPU, 122GB RAM. I am nowhere near that limit in usage.

I am unable to pass the 600k connections limit. Since there doesn't seem to be any O/S limitation (plenty of RAM/CPU/etc.) on either the client nor the server what is limiting me?

I have edited /etc/security/limits.conf as follows:

* soft  nofile  20000000
* hard  nofile  20000000

* soft  nproc  20000000
* hard  nproc  20000000

root  soft  nofile 20000000
root  hard  nofile 20000000

I have edited /etc/sysctl.conf as follows:

net.ipv4.ip_local_port_range = 1024 65535  
net.ipv4.tcp_tw_reuse = 1 
net.ipv4.tcp_mem = 5242880  5242880 5242880 
net.ipv4.tcp_tw_recycle = 1 
fs.file-max = 20000000 
fs.nr_open = 20000000 
net.ipv4.tcp_syncookies = 0

net.ipv4.tcp_max_syn_backlog = 10000 
net.ipv4.tcp_synack_retries = 3 
net.core.somaxconn=65536 
net.core.netdev_max_backlog=100000 
net.core.optmem_max = 20480000

For Apollo: export APOLLO_ULIMIT=20000000

For ActiveMQ:

ACTIVEMQ_OPTS="$ACTIVEMQ_OPTS -Dorg.apache.activemq.UseDedicatedTaskRunner=false"
ACTIVEMQ_OPTS_MEMORY="-Xms50G -Xmx115G"

I created 20 additional private addresses for eth0 on the client, then assigned them: ip addr add 11.22.33.44/24 dev eth0

I am FULLY aware of the 65k port limits which is why I did the above.

  • For ActiveMQ I got to: 574309
  • For Apollo I got to: 592891
  • For Rabbit I got to 90k but logging was awful and couldn't figure out what to do to go higher although I know its possible.
  • For Hive I got to trial limit of 1000. Awaiting a license
  • IBM wants to trade the cost of my house to use them - nah!
Backlash answered 30/3, 2015 at 23:52 Comment(3)
Can't really tell how to increase the throughput. However, checkout kafka.apache.org . Not sure about the MQTT support, but it seems capable of extrem throughput / # clients.Calvano
did you try mosquitto? (mosquitto.org)Rocha
Trying Hive, Apollo, Mosquito, Active, Rabbit, mosquitoBacklash
B
5

ANSWER: While doing this I realized that I had a misspelling in my client setting within /etc/sysctl.conf file for: net.ipv4.ip_local_port_range

I am now able to connect 956,591 MQTT clients to my Apollo server in 188sec.


More info: Trying to isolate if this is an O/S connection limitation or a Broker, I decided to write a simple Client/Server.

The server:

    Socket client = null;
    server = new ServerSocket(1884);
    while (true) {
        client = server.accept();
        clients.add(client);
    }

The Client:

    while (true) {
        InetAddress clientIPToBindTo = getNextClientVIP();
        Socket client = new Socket(hostname, 1884, clientIPToBindTo, 0);
        clients.add(client);
    }

With 21 IPs, I would expect 65535-1024*21 = 1354731 to be the boundary. In reality I am able to achieve 1231734

[root@ip ec2-user]# cat /proc/net/sockstat
sockets: used 1231734
TCP: inuse 5 orphan 0 tw 0 alloc 1231307 mem 2
UDP: inuse 4 mem 1
UDPLITE: inuse 0
RAW: inuse 0
FRAG: inuse 0 memory 0

So the socket/kernel/io stuff is worked out.

I am STILL unable to achieve this using any broker.

Again just after my client/server test this is the kernel settings.

Client:

[root@ip ec2-user]# sysctl -p
net.ipv4.ip_local_port_range = 1024     65535
net.ipv4.tcp_tw_reuse = 1
net.ipv4.tcp_mem = 5242880      5242880 15242880
net.ipv4.tcp_tw_recycle = 1
fs.file-max = 20000000
fs.nr_open = 20000000

[root@ip ec2-user]# cat /etc/security/limits.conf
* soft  nofile  2000000
* hard  nofile  2000000    
root  soft  nofile 2000000
root  hard  nofile 2000000

Server:

[root@ ec2-user]# sysctl -p
net.ipv4.tcp_tw_reuse = 1
net.ipv4.tcp_mem = 5242880      5242880 5242880
net.ipv4.tcp_tw_recycle = 1
fs.file-max = 20000000
fs.nr_open = 20000000
net.ipv4.tcp_syncookies = 0
net.ipv4.tcp_max_syn_backlog = 1000000
net.ipv4.tcp_synack_retries = 3
net.core.somaxconn = 65535
net.core.netdev_max_backlog = 1000000
net.core.optmem_max = 20480000
Backlash answered 1/4, 2015 at 20:4 Comment(3)
We tried with different platforms and irrespective of the CPU/RAM, we couldn't go beyond 20K connections in Apollo. We are running a test mqTT client (uses Paho library) to open connections. Any suggestions?Cleanlimbed
What we observe is beyond 20K connections, new connections are opened at a slow rate. The CPU usage, memory in server are under control. We also tried running the clients from different hosts. Same result.Cleanlimbed
@KarthikMurugan It's 2021, but were you able to solve this issue? Have the same use case and getting stuck at 2k connections.Fluidize

© 2022 - 2024 — McMap. All rights reserved.