Our Balance 20 has become really unstable. It seems to crash about once per day. The symptoms are that the web interface becomes unreachable and within a few minutes, it stops routing. A power cycle is required to resolve the issue.
I also see that the CPU usage is frequently > 80%.
I submitted a help ticket and they looked at our unit. The conclusion was that there are too many active sessions and that we’re overloading the router. This seems a little surprising.
We’re using this in a household with 6 people. We have 3 WAN interfaces (I bought the 3rd WAN license). The 3 interfaces have a total of 21Mbs downstream in the best possible case.
Only default firewall rules are on. We do have several Outbound routing rules. Maybe I should make sure that no extra switching is going through the Balance by putting everything on an external switch?
Maybe we need a Balance 1, but it just seems surprising that we can overload this in a home environment. I also need to balance 3 wired WAN interfaces.
Thanks,
Jonathan
if you go to Status > Active Sessions how many do you have when the CPU is high?
1 Like
The router just crashed after being up for 6hrs.
It looks like I had about 100-125 active sessions when this happened. The triggering event was that I changed the Outbound policies around a little. High CPU for about 1min, then the web UI became unreachable.
At this point, it seems like when I change anything there’s about a 50% chance the router will crash. I suspect the troubles began when I added WAN3 several months ago. It seems to be getting worse over time, which I don’t understand.
It would be nice if they had a watchdog that would reset the router when it goes out to lunch.
@jonashbrook … We had a similar experience some time ago. We thought one of the B20s we maintain was operating well within published specs. And, indeed it was. But, in the aggregate we taxed the CPU excessively (which we could see) and also the RAM (which we could not see). There were two WANs, two APs, two PepVPNs, a myriad of firewall rules and outbound policies – etc.
Fortunately, Peplink support was able to decode what was happening via RA and diag report. (Try that with a competing brand … 
) In simple terms, we had “too much going on” while not exceeding any one spec. The solution was to upgrade the customer to a B1 Core. The difference in performance was huge.
Incidentally, when you “lose” the UI for a period of time that’s likely to be an early indication of such issues. The web UI is prioritized behind core functions of the device.
One option you have (which I understand may not be a “complete solution”) is to set the router for a periodic reboot.
1 Like
I have an active ticket open with the same symptoms but on a Surf SOHO MK3 that has been running great for years. They have not diagnosed anything yet, but starting earlier this week the router freezes about every 12-24 hours, web interface unresponsive, internet access stopped. Only fix is a reboot. I suspect it’s because of all the IoT stuff that’s slowly been added, although typical number of active sessions is well under 100. (Usually about 80.) I have all of the IoT stuff on a wireless mesh whose connection to this router is doing NAT, so to the Pepwave all the IoT stuff looks like one LAN IP address. (Surf is recommended for <25 “users”.) Configuration has been stable for years, but a couple of new IoT devices were added to the mesh last week.
How do you set the router for a periodic reboot? I use schedules now for controlling other things, I just don’t see anyplace in firmware 8.0.2 to attached a reboot to a schedule. I was thinking of just putting the power adapter on a programmable timer that would cut power for 1 minute every 12 hours, but if I can get the router to reboot on a schedule, that might be less of a kludge.
Hi. I don’t have a SOHO close at hand but in most Peplink/Pepwave devices the support page can be reached at https://[device address];[port number]/cgi-bin/MANGA/support.cgi . Basically, just substitute the last part of the string when you have already logged on via the UI. Then, look for an option something like this:
I’ll invite a “SOHO expert” to correct what I’ve said if necessary.
You situation is soooo common! The situation I described (above) accrued exactly in the manner you stated – a slow but steady accrual of demands. (Oh yes … there were also three VLANs in use in my example situation.) System resources keep getting used up until – bam! – unpleasant things happen. The SOHO is a great box but everything has its limitations.
1 Like
Thanks, I’ll take a look. What it also underscores is that as homes & facilities slowly load up on IoT stuff, router makers are going to need to start upping the CPU & resource baselines (RAM & whatever) to handle that even on entry level devices. I’ve been considering upgrading to a Balance 20 in the hopes of gaining some horsepower if this is what is going on as I already offloaded all of the WiFi to two wireless meshes as they provide much better coverage, so I don’t need a router with WiFi. In that regard I wish Pepwave (& other router makers) did what NAS makers do and give you some idea of the hardware (chips, speed, RAM) between models whose functional specs look similar, so you potentially have a sense of their respective margins.
) In simple terms, we had “too much going on” while not exceeding any one spec. The solution was to upgrade the customer to a B1 Core. The difference in performance was huge.