Hi,
I'm trying to figure out what is going on.
After migrating whole network infrastructure to Mikrotik switches (CRS354 & CRS326) I am observing extremely slow DHCP performance.
It takes over 30-40 seconds to obtain an address - so long, that computers often initially report "no internet connection is available" and get the APIPA address (169.254.0.0/16) just to finally get the desired LAN address few seconds later.
The network itself works perfectly - pings between any nodes are minimal, transfer rates are going near GbE capabilities, even internet transfers are very high (three network operators, each one about 600-700 Mbps).
I was observing it in the simulated environment, when I connected new infrastructure to original - LANs and WANs connected to edge switch. Getting IP address from DHCP always took ages comparing to situation when I connected the device directly to any of the original LANs. I assumed that it might have been caused by some STP protocols on other switches (it was a mixed environment, mostly Netgear switches, some TPLinks, some HPs) but unfortunately after removing everything else there are no improvements.
I don't think it is the issue of DHCP server(s), after connecting directly to routers (some LANs are handled by Stormshield, some by Mikrotik router) DHCP works as charm. The network holds 6 LANs, all are well separated physically and with VLANs. Of course all connectivity between Mikrotik switches (4 CRS354 and 2 CRS326) goes via trunk SFP or QSFP links (each link contains only VLANs required in the network that is needed in specific location). No loops are created, all switches are set to MSTP protocol, I have checked LANs with sniffing to ensure that there are no leaks between LANs.
No DHCP snooping option is set on any of the switches - but I don't think that it would make any change (or maybe I am wrong?). No rogue DHCP servers are present (as observed and sniffed) - and even if there were any it is very unlikely to have them in all six LANs.
Any idea what might be wrong? Where should I start? The only thing I didn't check yet was sniffing exact DHCP procedure just at the DHCP server and behind first Mikrotik router - but after working on site for 19 hours while reconfiguring the network, replacing patchcords and checking if everything works as planned I wasn't able to spend there any more time
What would you do to determine the source of problem?
Greetings and Happy Easter to Everyone!
I'm trying to figure out what is going on.
After migrating whole network infrastructure to Mikrotik switches (CRS354 & CRS326) I am observing extremely slow DHCP performance.
It takes over 30-40 seconds to obtain an address - so long, that computers often initially report "no internet connection is available" and get the APIPA address (169.254.0.0/16) just to finally get the desired LAN address few seconds later.
The network itself works perfectly - pings between any nodes are minimal, transfer rates are going near GbE capabilities, even internet transfers are very high (three network operators, each one about 600-700 Mbps).
I was observing it in the simulated environment, when I connected new infrastructure to original - LANs and WANs connected to edge switch. Getting IP address from DHCP always took ages comparing to situation when I connected the device directly to any of the original LANs. I assumed that it might have been caused by some STP protocols on other switches (it was a mixed environment, mostly Netgear switches, some TPLinks, some HPs) but unfortunately after removing everything else there are no improvements.
I don't think it is the issue of DHCP server(s), after connecting directly to routers (some LANs are handled by Stormshield, some by Mikrotik router) DHCP works as charm. The network holds 6 LANs, all are well separated physically and with VLANs. Of course all connectivity between Mikrotik switches (4 CRS354 and 2 CRS326) goes via trunk SFP or QSFP links (each link contains only VLANs required in the network that is needed in specific location). No loops are created, all switches are set to MSTP protocol, I have checked LANs with sniffing to ensure that there are no leaks between LANs.
No DHCP snooping option is set on any of the switches - but I don't think that it would make any change (or maybe I am wrong?). No rogue DHCP servers are present (as observed and sniffed) - and even if there were any it is very unlikely to have them in all six LANs.
Any idea what might be wrong? Where should I start? The only thing I didn't check yet was sniffing exact DHCP procedure just at the DHCP server and behind first Mikrotik router - but after working on site for 19 hours while reconfiguring the network, replacing patchcords and checking if everything works as planned I wasn't able to spend there any more time
What would you do to determine the source of problem?
Greetings and Happy Easter to Everyone!
Statistics: Posted by joshuapl — Sat Mar 30, 2024 5:36 pm