RB751U-2HnD , 100% cpu usage, winbox login lag

hi guys..

need some help here..

i’ve recently bought a mikrotik RB751U-2HnD
and i’m getting some weird problems..

  1. winbox lag when trying to login
  2. high tx drop on certain interfaces..
  3. 100% CPU spike continues..
  4. “patterned” drop on my bandwidth/interface

i would like to first troubleshoot the 100% cpu SPIKE that im having…

refer attached pics..
mktikimg.png
if you refer to the bandwidth monitor on my pc you can see the drop is evenly spaced…
bwidth.png
which leads me to conlude that is a cycle of cpu spikes that cause a temp freeze or something like that

edit : learn a new trick , tools > profile to see utilization..
flash_highcpu.png
any idea why my “flash” is taking up so much memory ?

any info/data you guys need to help me to troubleshoot please help…

ps : i’ve returned an earlier unit, cause i thought it was some hardware fault and this is the second unit and its having the same symptoms

any idea why my “flash” is taking up so much memory ?

I’d like to know what flash is! I never saw that.

Saludos.

‘Flash’ process is writing to flash memory (NAND).
Check your settings about logging/graphing.
Something is permanently writing NAND, also check in Winbox
System/Resources/Total Writes.

HTH,

perhaps you should

/sys check-disk

the router will reboot once you apply it

well logging.. i only have the standard login and also a duplicate of some to forward it to a syslog server

graphing, well i dont use it, and there is no interface currently cfg to be graphed..

refer attached pics
loggraphs.GIF
regarding total writes, its quite high.. according to the screenshot i posted..

okie will try and and get back to you

thanks for helping

just did the /sys disk-check

and i’ve notice a “Weird” problem why has it rebooted and my int2 went to 10mbits mode ??
10int.png
10mbits.GIF
not sure if it has solve my cpu problem, will monitor/watch it again it the afternoon

a note on the 100% cpu usage, when my file(window) is winbox is open,

is the flash process suppose to spike to 100% and stay between 80 to 100% and stay there as long as the file(window) is open ?

fyi, when i open the file window, and it spikes to 100%, i automatically get disconnected from my winbox, is that suppose to happen ?guess /sys check-disk didnt do anything,
file_11.png

and i’ve notice a “Weird” problem why has it rebooted and my int2 went to 10mbits mode ??

There is problem with cable to ether2. Try to change it

the thing is.. my cables are connected to eth1, eth3, and etc5

eth2 is a master port with eth3 and eth4 its slave..

so there is no reason for to be in 10mbits mode at all…

still having occational spikes on the cpu usage for “flash” under profiling..

im guessing firmware bug and 5.12 having some bug…

@forces2010
Not connected master port always shows 10Mbit, half-duplex.
Your screenshot shows that you have over 120.000 NAND writes.
I dont know what is your RB total uptime but as some comparison,
RB750 which is running over two years has 3.500.000 writes.
As I wrote earlier something writes your NAND permanently, check your config.

okie.. any idea.. where section to check in particular ?

gonna do an export and run through it line by line…

edit : did that can’t seem to find a setting or config that would cause the high flash cpu usage

forces2010 - I am having a similar problem to you, which I have documented here - but basically when I use wifi and a skype call, I get high ‘flash’ usage, and I have not been able to get to the bottom of the issue.
I will be watching this thread with interest.

yeah.. through all my experience troubleshooting.. network/system/performance issues..

i got a STRONG feeling its a BUG…

hey guys just to check with you all, i came across this screenshot

why is his local bridge l2 mtu = 2290 same as his wifi…

but on my end screenshot post #1, my l2 mtu is 1594

packge.png
i’ve disabled all the other packages which i don’t use..

and im still getting high load on flash.. when i run through my export.. there is nothing that indicate

anything that would be using the flash…

any idea..

how much does it cost to log a case(this case) with mikrotik if i didnt buy from them ?

I’m currently using an RB750GL and was considering getting a RB751U-2HnD when I came across this thread. Has Mikrotik got back to any of you regarding this prob? I’m just afraid that I will end up with the same prob.

They got back to me to say they couldn’t reproduce the problem. I should try a different PSU (tried many with no difference) or a different unit (not easily possible for me) and then report back.

I’m using a separate (non mikrotik) wireless access point and having no troubles now.

Disable the ‘calea’, ‘ntp’ and ‘lcd’ packages, and see if the NAND problem goes away. Especially the ‘ntp’ package.

Do you have any information about why the NTP package could cause a problem? And any advice for keeping accurate time without it?

The ntp package is for installing an NTP server (and also a client) on your routerOS installation. This means the devices on your network can use your router to update their system clock.

If you disable it, routerOS will function only as an NTP client. Good enough if you want to use accurate clock for logging etc. The devices on your network should use external stratum 3 or 4 NTP servers like the pool.ntp.org clusters anyway.