RB750GL memory(ram) problem

I have two RB750GLs which are using up full memory the available in ust under a week.
The cannot run for a week without going to 94-100% memory and access to them via
ssh, telnet, mac or web is not possible.I have the older 750G and a 1200 running for
days with out any ram or cpu issues as deplayed by the 750GL
RB750GL_ram.png

In those wekely minute graphs, you can clearly see that memory is being emptied every day. what happens at that time?

I reboot it every other day to avoid lockup that happened over last weekend.
I had to drive there to power cycle it.

please make a support output file after reboot and then make another one when you see that lot of RAM is used. After that send both support output files to support@mikrotik.com

I just checked, its at 65.2% and climbing Is that high enough so that I can produce the
support output files? Am afraid waiting for it to go any higher will lockup it up like it
has done twice before.

I sent the support output files. As requested but I haven’t got any reply yet.

Sent from my GT-P7510 using Tapatalk

I also have this problem with X86 3ghz Intel with 512Mb of RAM. Every two weeks I reboot it to avoid problems. ROS is 5.14

ROS is 5.14

We have several new versions with improvements. Try v5.20 or v5.21 to see if your problem is still there. If you reboot it anyway, why can’t you upgrade?

I have been on v5.20 on a RB750GL and RB1200 since release and the problem still exists. The older hardware is not
affected like a 450g and 750g.

I can’t see the v5.21 on the download page. Is it really available?

Im having the same issue it seems. and yes i too have 493 that run fine. but the 750 lockup every few days. only thing i run on them is hotspot with about 20 users. if they are not power cycled they lockup. you can usually ping them but no other traffic

Am told its a driver issue being investigated.

Sent from my HTC Glacier

i have the same hardware (RB750GL), i run with an average of 30 users. i have the dhcp server running, load balancing and simple queues. i have been tested this hardware for about one week, and it crashes every 2 days. i’m gonna put more attention on the ram usage. thanks for the advisement.

It is not hardware related problem. I use API to make day/night bandwidth schemes. I found that if I turn off (disable) API, everything is working just fine. Problem solved.