Random mass timeouts after grading Debian.
#1

I'm having random timeouts several times an hour on 0.3.7 running Debian 9 on a dedicated server. All online players get timed out from the server at the same time. I didn't have any issues running Debian 8. I would revert back, but my server would then be open to the Meltdown exploit. I had this same issue when I tried running Ubuntu 16.04, which is what led to the switch to Debian. My network is in good condition and other services hosted from the same machine are not affected. Configuration is unchanged from Debian 8 to 9. Anyone have any suggestions to minimize or eliminate this issue?
Reply
#2

Are you using MySQL, if you do then which version of the plugin?
Reply
#3

What host are you using?
Reply
#4

The MySQL plugin is mysql-R41-2-Debian8-static (updated today to R41-4-Debian-Static, not likely to fix, but I'll let you know if it does)

We are using a dedicated server on Hetzner. Networking is on the higher end, and we had no issues before updating Debian 8 to 9.
Reply
#5

I can now confirm that updating the MySQL plugin did not help. Still getting timeouts. They do not coincide with any large numbers of MySQL queries or events in the dmesg log. Any other ideas?
Reply
#6

Are you sure the settings are the same, or do you assume they are the same?
Reply
#7

I've dug through the configs I've changed on the server and looked at change logs to see what default settings may have changed. Nothing different from the settings I had on Debian 8.
Reply
#8

Have you tried a second installation of SA-MP server, and trying Grand Larceny on it?

Have you tried clearing out the SA-MP stuff, and reinstalling it again?
Reply
#9

You might want to look at https://sampforum.blast.hk/showthread.php?tid=591057. I'm not sure that it'd help or not, but good luck.
Reply
#10

Quote:
Originally Posted by Logic_
View Post
You might want to look at https://sampforum.blast.hk/showthread.php?tid=591057. I'm not sure that it'd help or not, but good luck.
There's no solution to the issue in there though. Merely the fact that Debian 8 does fix it. Debian 9 seems to be on the borked side again.
Reply
#11

Quote:
Originally Posted by FufLa
View Post
There's no solution to the issue in there though. Merely the fact that Debian 8 does fix it. Debian 9 seems to be on the borked side again.
Bullshit... There's many avenues for investigation in there.
Reply
#12

I agree. Not sure where to start though. Can't seem to discern a trigger. Nothing from the kernel in dmesg, no uptick in traffic or SQL queries, no change in network availability or stability.

(Unrelated: I'm really cringing at the broken title of this thread.)
Reply
#13

I found what so far appears to be a solution (other than reinstalling Debian 8 ). This leaves the rest of your system intact.

Before using these instructions, you should be familiar with running a samp server on Linux as well as have a basic understanding of Linux and Debian.

Install debootstrap using your package manager (apt-get or aptitude for Debian).
Create a new directory with your choice of name. (I went with "/jessie-chroot" in the root filesystem)
Use debootstrap to install your choice of distribution (I went with jessie, which is the code name for Debian 8 )
As root or using sudo, use chroot to enter the new Debian environment you have installed.
Install libstdc++ 6 within your environment.
Make a new user to run your samp03server.
Install and run the samp03server. (I put mine in my new samp user's home directory)

This gives you the benefits of running your samp server from Debian 8, while allowing you the security of keeping an up to date Debian 9 as your primary system. (And for those of you who have already upgraded and are having timeouts, this will save you having to wipe and start over.)

These instructions should also work for Ubuntu, but I have not tested them and cannot confirm.
Reply
#14

I have a same problem when i upgrade to debian 9. I think, the best solution is to externalise service in a docker container from debian 8 image. This is a best practice if you want to stay updated for all services, for example you can create a docker container from php7.2-apache image for you website and create another container for database and another for your samp server. This makes it possible to configure each VM (docker container) with the perfect configuration for each service. Once the image (Dockerfile) created, you have only bound all the containers together. With this solution you can have last technologies php, mariadb on your website container and your samp server running without problems. It's not difficult to understand docker mechanic but just it's difficult to understand how create, build, start docker image and link container with the docker-compose but this is a best !!!!

Thank you for you post
Reply
#15

Look this for example :

https://github.com/Pyrax/samp-docker

This a Dockerfile for samp, it's really interesting and complete solution to this problem
Reply
#16

Has anyone even thought to check with Debian to see if there's an issue that they've actually found at some stage?
Reply
#17

Pretty sure a fairly big (once upon a time) roleplay server had this issue on Debian, moved to CentOS and problems were solved.

Quote:
Originally Posted by MustangV10
View Post
I've noticed this too on Debian 9, Ubuntu 18.04 and even CentOS 7. Running CentOS 6 (EOL 2020) and no problems whatsoever. Each time, stock configs with the exception of MariaDB over MySQL.
Not sure if they use 6 or 7 but there's no harm in just using 6, it's stable and works fine.
Reply


Forum Jump:


Users browsing this thread: 1 Guest(s)