X86 - very bad router performance after some time

RouterOS General
1

After some time, i can no longer login to my router with winbox. Then after some time, i can no longer login to ssh. Durring the period i actually can login to ssh, i can’t reboot the router - it just says:

[admin@BL-LUCKO] > sys rebo
Reboot, yes? [y/N]:
y
system will reboot shortly
error - contact MikroTik support and send a supout file (2)

Making the support file lasts very long (about 10 minutes - if i’m lucky

Theese are last entrys to log:

08:39:30 pptp,info TCP connection established from 31.45.146.50
08:39:30 pptp,ppp,info <pptp-0>: waiting for call...
08:39:31 pptp,ppp,info <pptp-0>: authenticated
08:39:32 pptp,ppp,info <pptp-0>: connected
08:39:32 pptp,ppp,info,account tsulek logged in, 172.16.0.253
08:39:32 pptp,ppp,info <pptp-tsulek>: using encoding - MPPE128 stateless
09:07:54 system,info,account user admin logged in from 93.138.27.40 via ssh
09:24:32 system,info,account user admin logged in from 93.138.27.40 via ssh
09:24:45 system,error,critical router was rebooted without proper shutdown
09:25:11 system,info,account user admin logged in from 93.138.27.40 via winbox
09:26:42 pptp,ppp,info,account tsulek logged out, 2831 2287443 14468255 23837 26947
09:26:42 pptp,ppp,info <pptp-tsulek>: terminating...
09:26:42 pptp,ppp,info <pptp-tsulek>: disconnected

Which is odd since after reboot, earlyer messages are lost.

2

what version of ROS?

have you sent supout to support@? what did they say?

3

We have this same problem. ROS 5.13. DELL Dual Core Server

Reboot about every 12 hours with the error - router was rebooted without proper shutdown.

I was going to upgrade to 5.14 tonight and then send in support stuff if it continued.

4

We actually think we narrowed down our problem. We were also having network wide speed issues. It turns out that the queues were the issue. I am not sure if its just x86 version doing this or not but basically we have a dell server running 5.13 dual NIC and are using it as a basic firewall for queues and redirects. As soon as we disabled all the queues the server stopped rebooting and the network traffic went back to normal.

If we turned on even one queue the whole thing turns to crap.

We still haven’t figured out why but at least the server isn’t rebooting. We just pushed the queues back out one hop for the time being.