Page 1 of 1
Dude on RB11004AHx4 Dude edition
Posted: Thu Apr 19, 2018 3:05 pm
by Insider
We are meeting quite bothering situation about Dude installed on, which we bought because of very poor behaviour on CCR1036, unfortunately it behaves here even worse and we need to restart system every 24 hours - but last weeks even that is not enough.. Any such experience ?
Re: Dude on RB11004AHx4 Dude edition
Posted: Fri Apr 20, 2018 5:53 pm
by maxsaf
Could you describe what is poor behavior in your case? We are also unhappy customers using a RB11004AHx4 Dude edition.
Re: Dude on RB11004AHx4 Dude edition
Posted: Tue Apr 24, 2018 9:42 am
by Insider
Could you describe what is poor behavior in your case? We are also unhappy customers using a RB11004AHx4 Dude edition.
After some time, usually after 14 day, on CCR the ROS started to colabse, first crashed dude it self, after came services starting to be not accessible (telnet, ssh, web), then finally was accessible only MAC telnet thru which it was possible to reboot device (once device passes traffic of several hundreds client it is big problem to wait until late night to restart it and get Dude working). Then there was gaps in graphs during regular operation.
The RB1100AHx4 is even worse, disappered gaps in graph, but systems fails with services sometime several times during day, sometimes after a week, so I made simple script, which restarts it every evening 2AM, even that doesn help. Router makes pure dude, nothing else. The fatality is that you have to access it physically to restart it so the double power unit is has no sense, once you need it to unplug both to get it working.. You can imagine what job is that in crowded rack, which is even 2km from office.. Thinking about returnig it..
Re: Dude on RB11004AHx4 Dude edition
Posted: Tue Apr 24, 2018 8:17 pm
by maxsaf
After migration to The Dude 6.41.2 on RB11004AHx4 Dude edition most of our graphs has gaps
Re: Dude on RB11004AHx4 Dude edition
Posted: Tue Apr 24, 2018 8:35 pm
by chechito
RB1100ahx4 is great for networking, very efficient and powerful machine
is not about stability, you must be aware of computational power of rb1100ahx4 router CPU, its a little A15 ARM quad core like a high end smartphone or tablet cpu, you must size the dude accordingly:
For rb750gr3
maximum of 40 services monitoring and maximum 4 days of raw value data storage and maximum 75 mega-byte database size
For rb1100ahx4
maximum of 80 services and maximum 4 days of raw value data storage and maximum 150 mega-byte database size
For ccr1036
maximum of 160 services and maximum 4 days of raw value data storage and maximum 300 mega-byte database size
Anything beyond that you will have not good results
beyond that you need a separate x86 device to host the dude
for big networks multiple virtualized instances of the dude will be needed, decent ssd storage etc etc
always keep the dude database size under 500mega-byte to keep good performance
Network monitoring IS NOT a light task
is a very intensive one not only in terms of CPU usage but in terms of storage i/o too
NEVER use onboard storage to sotrage the dude database for long term usage, in the case of rb1100ahx4 dude edition always use the 60gb ssd to host it
because monitoring is very write intensive it leads to a premature exhaustion of finite and scarce write cycles of the flash memory
Re: Dude on RB11004AHx4 Dude edition
Posted: Tue Apr 24, 2018 9:01 pm
by jarda
Very nicely written. But where did you take the sizing numbers from?
Re: Dude on RB11004AHx4 Dude edition
Posted: Tue Apr 24, 2018 10:48 pm
by Insider
After migration to The Dude 6.41.2 on RB11004AHx4 Dude edition most of our graphs has gaps
This is, what was making on CCR, no longer such problem on that RB1100AH4, only sometime, but it gets sorted out by averaging
Re: Dude on RB11004AHx4 Dude edition
Posted: Tue Apr 24, 2018 11:00 pm
by Insider
RB1100ahx4 is great for networking, very efficient and powerful machine
is not about stability, you must be aware of computational power of rb1100ahx4 router CPU, its a little A15 ARM quad core like a high end smartphone or tablet cpu, you must size the dude accordingly:
For rb750gr3
maximum of 40 services monitoring and maximum 4 days of raw value data storage and maximum 75 mega-byte database size
For rb1100ahx4
maximum of 80 services and maximum 4 days of raw value data storage and maximum 150 mega-byte database size
For ccr1036
maximum of 160 services and maximum 4 days of raw value data storage and maximum 300 mega-byte database size
Anything beyond that you will have not good results
beyond that you need a separate x86 device to host the dude
for big networks multiple virtualized instances of the dude will be needed, decent ssd storage etc etc
always keep the dude database size under 500mega-byte to keep good performance
Network monitoring IS NOT a light task
is a very intensive one not only in terms of CPU usage but in terms of storage i/o too
NEVER use onboard storage to sotrage the dude database for long term usage, in the case of rb1100ahx4 dude edition always use the 60gb ssd to host it
because monitoring is very write intensive it leads to a premature exhaustion of finite and scarce write cycles of the flash memory
Thank you for some hints, which I hadn't think of. The RAM will be probably crucial as system says 890MBytes used of 1GB total. This can be answer..
That is really funny, Dude just crashed completely..
Re: Dude on RB11004AHx4 Dude edition
Posted: Tue Apr 24, 2018 11:04 pm
by jarda
This happened to me three times already too. For the first time I tried to make mikrotik to find a reason and correct it. Unfortunately they were unable to do that. I solved that by importing a dude backup back.
Re: Dude on RB11004AHx4 Dude edition
Posted: Wed Apr 25, 2018 12:48 am
by chechito
Very nicely written. But where did you take the sizing numbers from?
that numbers come exclusively from my personal experience deploying the dude
Re: Dude on RB11004AHx4 Dude edition
Posted: Wed Apr 25, 2018 12:56 am
by chechito
RB1100ahx4 is great for networking, very efficient and powerful machine
is not about stability, you must be aware of computational power of rb1100ahx4 router CPU, its a little A15 ARM quad core like a high end smartphone or tablet cpu, you must size the dude accordingly:
For rb750gr3
maximum of 40 services monitoring and maximum 4 days of raw value data storage and maximum 75 mega-byte database size
For rb1100ahx4
maximum of 80 services and maximum 4 days of raw value data storage and maximum 150 mega-byte database size
For ccr1036
maximum of 160 services and maximum 4 days of raw value data storage and maximum 300 mega-byte database size
Anything beyond that you will have not good results
beyond that you need a separate x86 device to host the dude
for big networks multiple virtualized instances of the dude will be needed, decent ssd storage etc etc
always keep the dude database size under 500mega-byte to keep good performance
Network monitoring IS NOT a light task
is a very intensive one not only in terms of CPU usage but in terms of storage i/o too
NEVER use onboard storage to sotrage the dude database for long term usage, in the case of rb1100ahx4 dude edition always use the 60gb ssd to host it
because monitoring is very write intensive it leads to a premature exhaustion of finite and scarce write cycles of the flash memory
Thank you for some hints, which I hadn't think of. The RAM will be probably crucial as system says 890MBytes used of 1GB total. This can be answer..
That is really funny, Dude just crashed completely..
good point that i have not take in count
routeros in general do some strange things when runs out of ram memory
if your router have a heavy use of ram that's another resource the router will run out
every time i have used the dude on a router the router had idle resources enough to support the dude
in the case of a rb1100ahx4 with a configuration hungry of ram memory (surely with many simple queues) 1gb of ram gets used quickly
another frecuent problem with the dude is when you build a complex monitoring and later on you delete many elements to make it lighter, for some reason disk and ram usage does not come down even after a vacuum procedure
Re: Dude on RB11004AHx4 Dude edition
Posted: Wed Apr 25, 2018 12:58 am
by chechito
This happened to me three times already too. For the first time I tried to make mikrotik to find a reason and correct it. Unfortunately they were unable to do that. I solved that by importing a dude backup back.
ohh yeah
happened to me many times over the years, only needs a power outage to get the database corrupted
very important to do periodic backups with the dude, and get a reliable power source for equipment running the dude to mitigate the problem
Re: Dude on RB11004AHx4 Dude edition
Posted: Wed Apr 25, 2018 1:04 am
by maxsaf
Network monitoring IS NOT a light task
So why we have issues only with monitoring SNMP supporting devices and no issues using RouterOS API? Sure it is a same amount of data.