Magic dhcp server disable

We are facing a strange problem on an RB Metal 2SHPn in AP bridge mode
(with a single client in client bridge mode connected to it on WLAN interface)

We have configured a DHCP server on the AP bridge. This DHCP server seems to
deactivate itself on random occasions (on average once per day). For each
magic deactivation, the syslog contains an entry “DHCP server dhcp1 changed”
(in contrast, syslog says “DHCP server dhcp1 changed by admin” for intentional
changes via web interface or winbox).

Does anyone know a reason for such behavior and/or a fix/workaround?
(We frequently observed the problem with RouterOS 6.5. Yesterday we upgraded
to 6.6, and the some problem happened again at night.)

Thanks,
Hans

when there are no clients connected to wifi interface the interface will loose running flag, if interface is not running - DHCP-server becomes inactive.

Thanks for the response.

If this is the cause of the observed behavior, we need some way how to activate the DHCP-server again automatically as soon as a new client connects to wifi interface. This does not happen, but this is what I would expect from the DHCP-server. What currently happens is that the DHCP-server turns into a permanent “disabled” state. It shows in Italics in IP → DHCP Server, and the only way how to get it back to life is to log in with administrative permissions and enable it.

The problem does not occur each time there are no clients. “Usually”, the client can associate again to wifi interface after disconnecting and get an IP address. But sometimes, in situations we cannot reproduce, the DHCP-server remains disabled permanently until manual admin intervention.

please create and send support output file to support@mikrotik.com as it is not clear how and why exactly DHCP-server becomes disabled, it should become invalid, but when all the conditions are met - it should be fully active and should give out leases.

A short follow-up in this problem: There seem to be two situations in which the DHCP server magically turns into “disabled” state: (a) directly after an administrator accesses the router using WinBox, and (b) in other random situations. We do not know if these two issues are related or not.

For (a), we can repeatedly reproduce the problem. It is sufficient to start winbox, enter username and password of a specific administrator (and check “Load previous session”), and click “Connect”. As soon as WinBox connects, there will be log entries on the router (“user xxx logged in from yyy via winbox”, “dhcp server dhcp1 changed”), and the DHCP server is permanently down (until manual reactivation).

For (b), the problem occurs apparently randomly, and we are unable to reproduce it intentionally. We usually have one client associated with the wlan interface, but no further activity, and in the evening everything is fine, and in the next morning, the client does not get an IP address any more (we intentionally configured a rather short 1 hour DHCP lease time). Analyzing the log we can see a single “dhcp server dhcp1 changed” entry, without any other correlated events.

Removing the check mark on “Load previous session” in WinBox is a great workaround for probem (a) :smiley: , unfortunately not for (b)…