My Mikrotik server was shutting down about once every one or two weeks. I changed to a new PC, and still having the same problem. Anybody else seeing this? Version is 2.8.25. I am running pcq, web proxy and dns cache. Computer has AMD 1.4 GHz processor with 1024 MB of RAM. CPU load usually less than 5%, sometimes spiking up to 10-12%. I have one hard drive containing the O/S files and another for web proxy. No flash drive. Thanks!
Cameron
Is there a schedule doing it or maybe the whatchdog?
You are not clear on exactly what happens.
If it shuts off (power off) then it could be power management or cpu protection that shuts it off to cool the system down when it reaches near critical.
If it just hangs then it might be your hardware locking up. On IDE if a harddrive locks up, the entire system is left waiting for it to finish (which it wont if it locks) and your system appears frozen. If you somehow managed to unlock the harddrive the damage has already been done.
I suggest you do some actual investigating.
I get a kernel panic error. Like I said, the last router (that this one replaced) did this too. In fact, at this point it is my opinion that the legendary Linux reliability, at least as interpreted by Mikrotik, has become somewhat of a joke. I have eight Mikrotik routers running. Only one has run for close to a year without a reboot, but most of then shut down unexpectedly every several months. The worst one, is shutting down now every several days. Meanwhile, my Cisco routers have never had to be rebooted. Even my old NT router ran for 10 months without a reboot. In all fairness, the Mikrotiks are doing a lot more than simple routing, and I think that is part of the problem. It seems that there are now more features than the O/S can reliably support. Doing queues, web proxy, dns cache, routing, firewalls, etc is asking a lot of one machine. I think my latest problems started when I turned pcq on, so I am now turning it off and will think about using something else to do my shaping and web proxy duties. Too bad, because MT is a wonderful O/S. It just doesn’t seem to be reliable if asked to do much more than routing IMHO.
No reboots since I turned off pcq. Could there be something wrong with my configuration? Here are my mangle rules, mostly disabled now except for p2p. As you can see I have rules for p2p, pcq, voip and xbox. Am I just trying to do too much?
0 X in-interface=RoutedPublic p2p=!any action=accept
mark-flow=RoutedPublicQueueBypass
1 p2p=all-p2p action=passthrough mark-connection=p2p_con
2 in-interface=Public connection=p2p_con action=accept
mark-flow=p2p_down_limit
3 in-interface=Local connection=p2p_con action=accept
mark-flow=p2p_up_limit
4 X in-interface=Local dst-address=:3074 protocol=udp action=accept
mark-flow=xboxin
5 X src-address=:3074 in-interface=Public protocol=udp action=accept
mark-flow=xboxout
6 X src-address=:5004-5065 protocol=udp p2p=!any action=accept
mark-flow=VoIP-In mark-connection=VoIP-In set-tos=min-delay
3 in-interface=Local connection=p2p_con action=accept
mark-flow=p2p_up_limit
4 X in-interface=Local dst-address=:3074 protocol=udp action=accept
mark-flow=xboxin
5 X src-address=:3074 in-interface=Public protocol=udp action=accept
mark-flow=xboxout
6 X src-address=:5004-5065 protocol=udp p2p=!any action=accept
mark-flow=VoIP-In mark-connection=VoIP-In set-tos=min-delay
7 X dst-address=:5004-5065 protocol=udp p2p=!any action=accept
mark-flow=VoIP-Out mark-connection=VoIP-Out set-tos=min-delay
8 X p2p=!any action=passthrough mark-connection=pcq_down
9 X connection=pcq_down p2p=!any action=accept mark-flow=pcq_down_limit
Thanks for your assistance.
Cameron
It is very possible that you are experiencing resource starvation.
Resource starvation can easily result in crashes, reboots or trigger strange bugs.
Running out of memory is the typical one, however too many interrupts pr second can result in the unit “live-locking” (unit stops responding because it has to service interrupts) - this could crash a system in a very interresting way.
Could also be that you hit a bug in PCQ code 
More likely bug in Web-proxy
I still have web proxy running and no crashes since I disabled pcq. I have 1 GB RAM in this machine but I think maybe too many interrupts per second makes sense. My knowledge of this is limited but as I understand the problem actually gets worse as you add more memory. Maybe I would actually be better of with 512 MB?
I have this problem too and mostly because of not well tested hardware, and most likely MT support its board better than other more famous board which is not something blame anyone but I have 256 memory and it happened by no special trigger … could be … could be… could be… whatever
I personally think once it start working don’t play around with it and let it just do the wireless comm and setup your own server separately on linux box. for instance web proxy… NAT and blah blah.. unless you have not dedicate pc for that (LIKE ME) and you have no other choice (AT THE MOMENT)
hope it helps. thanks