Kernel support for server-side tcp fast open remains disabled

[snorkel]

Maybe @BrightSilence has an ideea. He runs Synology also.

[Alexey]

Two new blueprints/design drafts seeking feedback: Replacing TLS with Noise and TCP_FASTOPEN

It isn’t overwritten with every update, but with larger updates it can be. DSM 7.2 is coming up, which is a larger feature update (that also change the docker implementation to a new “container manager” interface. So reason to be careful with this update to begin with. But chances are this one will impact that file as well. I’d say just keep an eye on it after any big update. Or switch to a startup script that you can manage in the DSM interface itself to set these settings. That way you know i…

[snorkel]

DSM (7.1.1-42962 Update 4) and Docker (20.10.3-1308). I’m on these versions since a few moths ago.

[snorkel]

After NAS restart, the docker exec dosen’t show 3 anymore, only the error.

[snorkel]

The log shows this at node start:

INFO server kernel support for tcp fast open unknown {"Process": "storagenode"}

It runs without the tcp_fastopen parameter in docker run command. Maybe I have this problem because I run the node as root?

[BrightSilence]

@snorkel

Maybe @BrightSilence has an ideea. It runs Synology also.

It?

I’m not at my system atm. I will check some things tomorrow. I have the setting set on the host machine using a scheduled script at boot. Haven’t set anything in the docker run, but I’ll try to verify if that’s needed. Did you set TCP fast open to 3 before running your Docker container after restart?

[snorkel]

@BrightSilence
I was doing too many things for a free Sunday. Sorry… I corrected the mistake. When you are non native english speaker, it happens. … and with the speed at wich AI is developing, who knows in a few years who whould be at the orher end of the line .
Yes, I set the paramenter in the boot script, also in the sysctrl, and restarted. I checked it with that command and was showing 3.
Then stop container > remove > launch new container. And error… now is running without that parameter and it says that the state of tcp fastopen is unknown. I can’t tell if it’s working or not. That’s why it should have an indicator on the dashboard like QUICK has.
Thanks for any suggestions!

[Dunc4n1d4h0]

docker exec -it  storagenode cat /proc/sys/net/ipv4/tcp_fastopen
3

2023-04-03T18:44:20.577Z        INFO    server  existing kernel support for server-side tcp fast open detected  {"Process": "storagenode"}

Big thanks for idea to include --sysctl in docker run

Now… Does it change anything other than this message?

[waistcoat]

Now… Does it change anything other than this message?

Not that I’ve noticed yet.

netstat -s|grep Fast

Should give some output when a TFO connection has been made. I’ve done some tests and my router does pass TFO through correctly so I can only assume my node hasn’t seen a TFO request just yet.

[Roberto]

Perhaps it would be appropriate to put a notice on the dashboard for the Fast Open. As for the Quick

[Walter1]

Thans alot for your command.

Can you explain why you put

	--storage2.piece-scan-on-startup=false

into the command? I’m having this in the config.yaml.

Can you also explain what

	--log-opt max-file=5 \

does?

Thanks and kind regards,

[snorkel]

@Walter1
The first one, if it’s false, stops the File Walker from running. If it’s true, FW runs at node start, and after any updates and restarts.
See post “Tuning the filewalker”.
The second one reffers to logs; it’s a log rotate mechanism. One line specify the max size of one file, and the other the number of files that are created and retained. New files created after this value will delete the old ones, to keep the number of log files at that value. A new log file is created when the max size is reached.
If I remember corectly, this only works if you didn’t changed the location of the log files from default.
This is the expanded version:

My docker run commands for multinodes on Synology NAS

These are all the useful parameters that one storagenode can use, and the modifications that I use for the second node. Fill free to use them or not, is up to you. I put them here fore easy reference. These work for Synology NAS and I run them in sudo su mode. There is also the --user $(id -u):$(id -g) parameter, but I started without it. It seems that dosen’t work on Synology NAS. NODE 1, MACHINE 1: Run only once at install of first node: echo "net.core.rmem_max=2500000" >> /etc/sysctl.co…

Take notice that I keep FW off for 2 nodes that are on a NAS with only 1GB RAM. For other, I have 10GB and 18GB RAM and FW is on.
There are also some commands that reffers to how data is buffered and write on disks, to reduce the fragmentation. I keep those as default on systems with low RAM. I use different Docker ports for each node, because I was getting Quick missconfigured when I was using the same port. Now I don’t get Quick error ever.