my SXT 5HPnD has been running for 36 days, now the CPU suddenly goes to 100% and when i open profile it shows:
any idea what’s wrong? 
thank you
I have similar experience with 2 rb750gl. Every few days flash consumes 100% and only a reboot resolves it.
Am still looking for a solution.
Sent from my GT-P7510 using Tapatalk
signaling mod 
Sounds like the problem we are having on RB750GL, RB750UP and RB2011 
We have a ticket open but response from Mikrotik is slow.
yep. I also had a RB750UP which I replaced thinking its caused by its low memory with a RB750GL.
Turned out is the replace ment has the same issue.
Can you try to disable some functionality and see if it helps? Disable SNMP, graphs, logging to disk
For us, we only have snmp enabled. We have swapped the troublesome devices to RB1200’s for now. I will set up a few of the MIPS-BE devices next week and test disabling SNMP
I tried disabling SNMP but it didn’t help. I log to memory, do I need to disable that too?
Have you got a solution to this problem?
I had a similar issue recently with a Groove with 5.20. Luckily I managed to reboot it remotely and when it came up I loaded 5.21 onto it. Since then it’s been resolved.
The same thing happened to me just moments ago with an RB750UP (running 5.21) that is in the field. I was running winbox, and looking at the profiler at the time. All was looking good, then suddenly the CPU pinned at 100%, the profiler indicated massive flash activity, and before I could tell it to reboot, winbox disconnected. The 750 still responds to pings–and even passes bridged traffic through to devices behind it–but telnet, ssh and winbox time out, so I can’t tell it to reboot. Most likely, any other tasks that require CPU attention (DHCP et. al.) are not working either, but I am not on-site to confirm.
It is clearly “trying” to respond to outside requests*, but I’m guessing the CPU is so busy with whatever bug it is chewing on, that these attempts time out before they can accomplish anything. Hopefully I can get into the location tomorrow and physically power-cycle the box. In the meantime, I will keep trying to send reboot commands via ssh, and hope that one of them finds enough cycles to get through before giving up.
\
- For example:
telnet 10.0.2.150
Trying 10.0.2.150…
Connected to 10.0.2.150.
Escape character is ‘^]’.
[~20 seconds elapse]
MikroTik v5.21
[~20 seconds elapse]
could not receive confirmation to CTRL_READY from loader
— nv::message --------
Connection closed by foreign host.