Hi guys,
I have an ISP and cable operator, something very re you strange is happening in one sector of the network. Everything is working fine, with good SNR values, and suddenly all the CMs disconnect from the CMTS, the stay down for a few minutes (5-15 minutes), and then all of them come back with good SNR values again. This is happening every 2-3 days between 8pm and 10pm. I suspect of interference, but the strange thing is that it happens randomly. What could be causing this? Some is trying to hack me? If yes, what kind of device are they using? Trying to determine how to find the interference source.
Thanks
sounds like Downstream noise. Usually if its upstream noise the modem stays connected and you can see the noise level. 8 to 10 pm..... Thats typically when temps drop which will cause overdrive if your plant is not balanced correctly. i would check your MER on your internet Downstream channel. If its below 34 i would say thats your issue. then just keep back tracking until your MER gets better
I gonna say its downstream noise issue. Modem typically don't disconnect from CMTS if its upstream noise. So you would be able to see active noise level on the return. The first thing i would check is the Quality of your QAM channel running your internet service. Also are they losing TV services or just internet service. Alot of things can be wrong. If your internet channel in the lower frequency are the lower TV channels fussy? My guess since you said between 8pm and 10 pm that usually a big temp drop during that time so i would make your balanced correctly and not creating overdrive. could be AGC freaking out on a amplifier. You can usually tell that by looking at your MER on your digital meter. I would check that at customers house if its about 34 your downstream channel is probably not the issue. If it below that i would just work back until it shows a good MER then start from there.
what does the cable plant look like? example it's a small coax system with common downstream (broadcast, not narrowcast). each node has it's own receiver on the cmts or all are combined and then split? it would help to know how the signal leaves and comes back.
I'm asking because Way Way back... I had an issue that only happened on thursday evening from 8 to 9 pm. took a long time track down, but it was the generator at the dpw that would run on a self test schedule and it had a bad ground causing impulse noise on the return. rlejeune does have some good points, what I'm trying to say is I or we would need more information.
could be as simple as an issue with the dhcp server with lease time, arp, routing issues.
1. number of total customers and number customers that go offline
2. coax leaving the headend or fiber; how is everything combined
3. cmts configured with spectrum management? (example; snr is at 32 with 64Qam on the return. if snr drops to 22 mod profile goes from 64Qam to 16Qam)
4. increase lease time on modem from dhcp server. I recommend a month instead of a week. it's a private address, no need to have low lease time for modems, just for pc's or routers.
5. return signal generator at the termination point, the 11 tap at the end of line for example. carrier good when modems are gone?
6. can you post log files from the cmts?
7. re-reading your post, you mention one sector.. if you swap the downstream/upstream that feeds that sector with another that doesn't go down, does the issue follow or does the new sector now have the issue? (would show if cmts or cable plant because new sector would go down if cmts. old sector still going down would point to cable plant issue, like power supply for example)
I know more questions than a simple answer, but in most cases there isn't a simple answer. answers to questions above would paint a better picture of what you have to work with and we can give options
Hello
Yes dhcp is a good idea: you could check syslog if cm didn’t reboot
And yes context and configuration may help
If you are able to try easydocsis.com it will be easier to understand, check and share your issue
In any case good luck! Such problems are always difficult to solve
Dont forget to check the DOCSIS event logs from some of the affected modems
You can fetch it via SNMP
Have a look and see if any clues at time of dropout, eg T3 or T4 timeouts