Out of memory?

[Kennebec]

I could really use some advice. I have a clean install done last weekend on a Ubuntu 20.04.2 server. The machine has 32GB of RAM and is running a login server. The machine is only available on my LAN to my Windows 10 computer. after using it for an hour Linux has a message:

Akka's Linux Server Launcher
World: UP Zones: (10/30) UCS: UP Queryserv: UP
kennebec@mainelyadventures:/home/eqemu/server$ Out of memory!

I run top and it shows 29GB free.

top - 18:17:21 up 12:20, 1 user, load average: 0.19, 0.23, 0.16
Tasks: 193 total, 2 running, 191 sleeping, 0 stopped, 0 zombie
%Cpu(s): 0.5 us, 0.4 sy, 0.0 ni, 99.1 id, 0.0 wa, 0.0 hi, 0.1 si, 0.0 st
MiB Mem : 32000.2 total, 29979.6 free, 732.2 used, 1288.5 buff/cache
MiB Swap: 8192.0 total, 8192.0 free, 0.0 used. 30850.7 avail Mem


Pretty soon after that my client crashes if I don't logout the client and stop/start the server.

Is there a tweak I need to make to increase memory usage? The install was clean and smooth with no errors and all that is running on the machine is EQEmu.

Any advice is greatly appreciated.

[Kennebec]

Still hoping I can get some help on this. I increased the RAM to 64GB and did a clean reinstall of EQEMU on 9/24/2021. Running Ubuntu 20.04.3 and MariaDB 10.3.31. Started the server and logged in 1 client. Killed rats in Qeynos hills for 45 minutes until I get "Out of memory!" popping up on the Linux console.

free -m shows:
kennebec@mainelyadventures:/home/eqemu/server$ free -m
total used free shared buff/cache available
Mem: 64208 1383 60867 4 1956 62178
Swap: 40959 0 40959

Clearly not out of memory. I am not a Linux guru so not sure where to check next. I did a grep -r "Out of memory!" on /var/log and the only entry is the auth log for entering the command. I also reviewed the logs in /home/eqemu/server/logs and cannot find any errors.

Anyone have any clues for where I should check next? Thanks in advance.

[Kennebec]

Adding one more input that I was able to catch. When I get the "Out of memory!" message I logout of the client then run the "sudo -u eqemu ./server_stop.sh" command.

I am using the standard eqemu user eqemu password combo from the standard install

I received this error message tonight after one hour of play with a single character in a single zone killing rats:

Akka's Linux Server Launcher
Shutting server down...
Out of memory!
[sudo] password for eqemu:
gdb: unable to read password: Input/output error
Server shutdown...

[Kennebec]

Adding to the knowledge base on this issue. I started digging into the system logs and in /var/log/mysql/error.log I found this error at a time that coincides with the "Out of memory!" message:

2021-10-04 18:39:42 13 [Warning] Aborted connection 13 to db: 'eqemu' user: 'eqemu' host: 'localhost' (Got an error reading communication packets)

Researching that message suggests a tweak to the 50-server.cnf config file. so I changed max_allowed_packet to 256M. From what I read it defaults to 1M if not specified. I'll follow up with results, so far I have passed the 45 minute mark which isn't conclusive yet.

[Kennebec]

coming up on 2 hours now and still no error message. Posting this as a tentative resolution:

cd /etc/mysql/mariadb.conf.d
Sudo nano 50-server.cnf

find the line with #max_allowed_packet = xxM and remove the # and change the value so that the line reads:

max_allowed_packet = 256M

Save the file then reboot the server or restart mysql

I hope that saves someone some frustration. I'm (tentatively) happy to put this to rest.

[Kennebec]

Adding a final comment in case this comes up in someone's future search. I wiped the machine and did a fresh install of Debian instead of Ubuntu then did a clean install of eqemu and have been running without any memory errors and have been able to leave the server up for weeks at a time. As someone posted in another thread Ubuntu 18 was the last clean version for an EQEMU install.

[Hurb]

great thread.

i also used Debian 10 linux on an Amazon Web service EC2 virtual instance, and that has had continual up time and is public facing with 4 gb ram (upgradable to any needed number of cores/RAM!) .. it would be fun to stress test the server since I got $300 in AWS credits and could really see what happens with multiple users on different tiers, and even compile new servers (ubuntu / windows server)

Hurbquest http://www.eqemulator.org/index.php?...o&worldid=3264

[markusdabrave]

Thanks for this. I had the same issue ad this fixed it.

[Hurb]

i set virtual ec2 instance back to t3.small, 2 cpu /2gb debian 10 .. still good to go but then again i haven't tested more than a handful of users