internet connection check

by skynetbbs on

Every time I have a problem with internet connectivity (ADSL session timeout or a router inbetween that fails) the Flukso v1/v2 stops sending data to the Flukso server...removing the utp cable from the flukso lan port doesn't help.
I have to reboot and thus loose days of data which is probably still residing in the flukso cache?

According to Flukso website the heartbeat is correct : 28/03/2012
and an uptime of 3days 21hours

but no data is shown of it's 2 optical and 3 amp clamps???

last data on the flukso website 24/03/2012 :(

‹ Wrong time hour change First installation flm01 - sensors problems and questions v01 ›
icarus75's picture
icarus75 |

We've narrowed this down to an NTP issue on Belgacom Bbox2's. TBI.

icarus75's picture
icarus75 |

With commit 820cba a cronjob will check every 15mins whether the ntpclient daemon is still up and running. If not, it'll kick it back to life. Will be rolled into r216.

stefaang's picture
stefaang |

Hi Bart,
I think I'm having the same issue - no data has been updated on my dash since this morning, nor has PVoutput.org updated the info on my solar-panels-

I don't use a BBox2, but a FritzBox7390 as a modem, however i'm a Belgacom customer.
My system time seems to be in the 70's :-)
I haven't done a reset yet, if you want you can have a peek.

icarus75's picture
icarus75 |

Thanks for reporting the issue. Here's the relevant syslog trace from your FLM:

  1. Jan  1 00:01:09 flukso-79a117 user.info sysinit: 0.openwrt.pool.ntp.org: Unknown host
  2. Jan  1 00:01:39 flukso-79a117 user.info sysinit: 1.openwrt.pool.ntp.org: Unknown host
  3. Jan  1 00:02:06 flukso-79a117 user.err heartbeat[837]: POST 'https://api.flukso.net/device/abc': -2
  4. Jan  1 00:02:06 flukso-79a117 user.err heartbeat[837]: Name or service not known
  5. Jan  1 00:02:09 flukso-79a117 user.info sysinit: 2.openwrt.pool.ntp.org: Unknown host
  6. Jan  1 00:02:39 flukso-79a117 user.info sysinit: 3.openwrt.pool.ntp.org: Unknown host

It shows ntpclient trying to connect to a different ntp server pool every 30secs. After 4 unsuccessful attempts, ntpclient gives up and terminates. The connection problem is DNS related as the POST to api.flukso.net failed as well with error 'Name or service not known'.

Later heartbeats succeeded. So the DNS issue seems to resolve itself after some unspecified time. This confirms that the cronjob check of 820cba will do the trick.

I've now kicked the ntpclient back to life on your FLM.

skynetbbs's picture
skynetbbs |

When a PPPOE session is renewed;

i can check in the logs of 192.168.1.1 (BBOX) that it's date is also in 2003 instead of 2012 for time being.

The 192.168.1.1 broadcasts itself as DNS server in it's DHCP leases... this DNS is flushed at this moment and an "nslookup" fails ... for at least 5 minutes give or take.

personally I'd use 8.8.8.8 or 208.67.220.220 as dns server

PPPOE sessions take 96h before session is terminated at Belgacom but recently it changed to 48h I believe