Services watchdog - feature request ("reset states" checkbox)
-
Dear developpers.
Can you please add possibility to reset states before starting died service.2.3.3-RELEASE-p1
WAN (pppoe).
When connection resets (once per 2-3 days) it gets new ip address. OpenVpn services crashed at that momet. Service Watchdog correctly starts it but some IP phones hanged with wrong states.So, if Watchdog could reset states, it would be perfect.
Something like this (see. attachment).
-
This just doesn't make any sense in this package. All this package does is adding a cronjob that will restart stopped service (and optionally send a notification). What kind of "reset states" you imagine with things like DNS server, DHCP server, NTPd or whatever?
-
That is why i ask it like option.
For crashed open vpn service i have to manually reset states in order to force voip-sip device to reregister.
I received notification, hence i need to reset states, so why not to do it automatically? -
I think you've missed the point - that option makes no sense for majority of services.
I have no clue what's wrong with your SIP, but you'd better get it fixed with the devices manufacturer or in relevant OpenVPN code in pfSense.
-
It's the old problem in pfsense with VOIP, and not only mine.
For ex. starting 2.1 there is the new checkbox: "State Killing on Gateway Failure" (System/Advanced/Miscellaneous) although you could say "no sense for majority of wan setups. you'd better get it fixed with the ISP".
And this feature (reset state table) helps in case when it's detected on WAN connection level.Now discovered another problem: some services sometimes crashed and pfsense (not SIP device) opens or keeps wrong state by-passing openvpn channel.
Service watch dog restores services, fine, it really helps. But needs some additional improvements. -
This doesn't go anywhere. Having a WTF checkbox confusing users there because you have a fringe issue with something completely different that needs to be fixed in a different place… Say, someone else has another issue somewhere that requires restarting DNS server. So, every service there will have a checkbox to restart DNS, just because they have this issue? Huh...
-
It's a not a single issue.
Many users with pppoe have this problem.
Well, i can play with loss threshold parameters in order to force fire "wan disconnect", so probably state tables will be reset by "State Killing on Gateway Failure". I can use "aggressive" optimization to flush idle states more quickly. But most reliable approach to make reset on specific event.And why confused? It has clear explanation.
Many users have similar problems.
Simply not reported or solved by reboot/unplug-plug/smth. else. -
And why confused? It has clear explanation.
Geeeez. Because you don't add a global checkbox to things where it makes zero sense. It's like putting a defrost button on your toilet because it makes sense with your microwave. Get the bug fixed properly, this ain't a place to hack around it.
If you want a custom code to be run on service restart which could be optionally configured for individual services (as opposed to global WTF checkboxes) I'm sure such pull request would be welcome and acceptable.
-
problem solved.
decreased "time period" (Time period in milliseconds over which results are averaged. Default is 60000.) to 5000
Now wan correctly goes down, OpenVpn do not crash.