On May 6, 2015 10:05 AM, "Rod Roark" <[email protected]> wrote: > > Here's a puzzle for you experienced network administrators. > > My mom-in-law is in a nursing home and I wanted to set up a MagicJack > for her to save some bucks on her phone bills. The facility has Wi-Fi > with unsecured guest access where you just have to accept the terms of > use via a captive web portal. > > Long story short, I set up an old netbook computer for her with Linux > Mint Debian Edition, and disabled Network Manager and configured wlan0 > in /etc/network/interfaces. I also wrote a PHP script that runs every 5 > minutes via cron to automate the portal login when required. It also > detects when Internet connectivity is lost and executes an ifdown and > ifup of > wlan0 in that case. > > The problem: Frequent wi-fi outages, evidenced by DNS lookup failure. > Clues are: > > 1. They only happen during the day when lots of staff or visitors are > around. The exception is shortly before midnight every night, which I > figure is by design. > > 2. They occur at random times, about 6 times per day. Thus clearly not > an intentional timeout. > > 3. They are cured by the ifdown/ifup cycle. If that doesn't happen then > the connection stays lost. Doing it fixes the problem every time. Thus > it has nothing to do with maxing out bandwidth. > > 4. The netbook is always assigned the same IP address by DHCP. Thus it > can't be IP address conflict (DHCP server misconfiguration). > > 5. It appears there are multiple access points in the facility with the > same SSID. > > The only possible causes I can come up with are: > > (a) Someone is rebooting access points when they think that might fix > something. This seems somewhat unlikely because it's also happening on > weekends when administrative staff are not around. > > (b) Access point malfunction under heavy use. > > Any other ideas? > > _______________________________________________ > vox-tech mailing list > [email protected] > http://lists.lugod.org/mailman/listinfo/vox-tech
I wouldn't rule out IP address issues. It's possible that the AP / DHCP server / etc. is being slammed with too many DHCP leases, resulting in address exhaustion for that particular subnet. This is especially prevalent when a small business tries to use a residential or SOHO product to run their wifi without moving from the default 192.168.1.100-255 range (which is the most common on SOHO and residential routers), but I've seen it happen even at hospitals with high-end Cisco-everything networks. -- Ryan S. Northrup
_______________________________________________ vox-tech mailing list [email protected] http://lists.lugod.org/mailman/listinfo/vox-tech
