Do you restart your router / how often?

Best to seek clinical help if you suffer from murophobia, rebooting your router ain't gonna do squat.

3 Likes

First things first:
First compromise is a workstation in your LAN. This is game over as you can access all devices in that network. Unless you have strict VLAN/zoning with firewall/routing inbetween, zuorat or not, is a moot point.
So please explain to me why this wouldnt be your first thing to fix?
Second: if it managed to push zuorat to your soho router on day 1, why wouldnt it be able to do it the minute after you reboot the soho router with the same firmware/bugs?

I noted: pc perspective is not nuanced enough and i will not trust their writing in the future. It seems to try to cause panic for all the wrong reasons

(My job is being a CISO, so i do think i know what i'm talking about)

1 Like

TY !!!!!!!!

no , i have never restarted it. But i did notice sometimes it does restart itself according to the logs but after turning off the wifi, it doesn't do that anymore.

It’s not hard to understand but you are speaking/writing to the OP in a very demeaning way.

2 Likes

You're right. I'm sorry for that. I just get frustrated sometimes and get carried away.

3 Likes

Sometimes. Occasionally I find the DL speed has halved, restarting the interfaces etc doesn't fix the issue. I've no idea of the cause but a reboot gets it going.

EDIT: This got me to a bit more digging and found softirq was hitting 100%, turned on SW offloading and it looks as if my issue of DL speed dropping after some days is resolved. Obviously I need to monitor for a while but I am back to 500/500 or real close from 200/500.

I actually restart my router every 24 hours.
Sometimes the Ethernet switch just completely drops out, sometimes my 4G Stick is freezing up, sometimes the 2.4G WLAN is kicking people out for no reason... all solved by a simple reboot.
And since I have a home server to which I want to have access at all times, having an automated reboot routine every 24 hours means that if I lose access to the server for some reason I will regain it again after maximum 24 hours without external intervention (I'm 6 months away from home).
The reboot script has actually saved me twice now.

Ow my god!!! This is a disaster!!! These last 2 posts. Honestly... How can you live with this unstability?! Have you all filed bugreports? Are you sure your hardware is still OK?
When i'm in my lazy period, i have uptimes of a year and more! No issues whatsoever.

Are there people on this forum that can help you fix this?

1 Like

Not sure whether running a router for 12 months straight without security updates is that much better than rebooting it every day...

I agree though that as a general stabilization method it seems sub-optimal. On the other hand if flaky hardware will still be usable that way, why not?

I restart it daily because I stop the Wifi interface at night (9 p.m.) with a cron job but did not find a way that worked to bring it up in the morning so I just restart it in the morning at 7 a.m.. Works for me as the router is for home purposes (not business or something).

Honestly, there are not many remote exploit kernel vulnerabilities in the linux kernel. So not rebooting is not such a big deal, security wise.

Next to that: i did not say i dont update packages! But since i run on x86 hardware, there is no "good" way to upgrade openwrt... So just some opkg magic does the trick

Not sure that in OpenWrt you can replace the C-library for all running programs at run-time... Not that embedded clibs see that much attention, so I would assume a good fraction of exploits stay hidden, and that can not be resolved by updates because bugs need to be known and fixed first.

Hrm, not sure that opkg is a full fledged package manger for such operation. I would probably tend, even with an x86 router to "flash" full builds more often than every 12 months, but hey everybody has their own choices... (which also applies to rebooting strategies I guess).

You can't, the libc itself isn't packaged (only a dummy package for dependency tracking) - and literally every package hard-depends on the libc it has been built against (musl, in almost all cases).

1 Like

I reboot my WRT3200ACM once a month to see how much bandwidth I used in a month.

Mmh, so save yourself one subtration (with the added challenge of taking care of potential counter wrap around). Obviously a personal choice, but I would guess the calculation taking less time than a reboot.

I only reboot if installing new firmware build.
This is imo one of the features of OpenWrt, that it runs and runs and r….
Only thing is, if you run on HW where only unstable drivers are available and a service restart is not enough. But even then i try to switch to „stable“ hardware.
I currently running trunk with AUC and update the firmware around every half year. Takes about 5mins.

1 Like

Same here. Albeit the same is true for my other Linux hosts, except these see considerably more kernel/important library updates and hence reboots.

But I accept that there are use-cases where more frequent restarts are acceptable/desirable.

Yes, however if an otherwise capable enough piece of hardware can be saved from the land-fill by more frequent reboots, I can see that as a decent justification for going the "reboot early and often" route.

Kernel upgrades require a reboot, the rest does not. If you have libraries that changed, you can quite easily use the new ones by reloading. They are dynamically linked for a reason.
On the landfill and stuff... I agree, but only partially. If stuff is broken, you dont know what else will go wrong. So a reboot might save you now... But you will probably be in deeper sh*t quite quickly.

At that point all I say, I was under the same impression in the past, but that is not how things work. For shared libraries that are loaded and used, updating the library will not affect the loaded instances IIRC. In short there are systems that can be fully upgraded (sans kernel) without a reboot, default OpenWrt is IMHO not one of these systems.

If stuff is not apparently broken already you still "don't know what else will go wrong"... this is not a good argument, IMHO.