[HamWAN PSDR] Haystack Site Outage
Chris S. Munz-Michielin
christopher at ve7alb.ca
Tue Mar 14 10:28:20 PDT 2017
For what it's worth the Victoria uplink is a 30/5 business cable
connection which VPNs back to the Seattle edge routers. No BGP
connectivity up here at present, but we MAY have something coming down
the pipe via a Vancover datacenter after we build out Saltspring Island
(hopefully) this summer.
Chris
On 3/14/2017 2:34 AM, Bart Kus wrote:
> I'm not sure that's a fair assessment. The link to SnoDEM is only
> -77dBm and only on hpol (vpol is -90 ish). Here's the perf that link
> gets:
>
> [eo at SnoDEM.Haystack] /interface wireless registration-table> /tool
> bandwidth-test 44.24.242.6 direction=receive
> status: running
> duration: 30s
> rx-current: 4.4Mbps
> rx-10-second-average: 4.7Mbps
> rx-total-average: 4.4Mbps
> lost-packets: 478
> random-data: no
> direction: receive
> rx-size: 1500
>
> [eo at SnoDEM.Haystack] /interface wireless registration-table> /tool
> bandwidth-test 44.24.242.6 direction=transmit
> status: running
> duration: 30s
> tx-current: 11.9kbps
> tx-10-second-average: 27.6kbps
> tx-total-average: 26.4kbps
> random-data: no
> direction: transmit
> tx-size: 1500
>
> That TX path makes the link useless. Better to turn it off! Or maybe
> ditch the weak polarity.
>
> Then there's Nigel's + my link (I'm NOT uplinking right now), both
> sitting @ -81dBm, which is pretty low. Or at least we were. I see
> Nigel's @ -68dBm right now. I don't know how to explain that. Here's
> the link speed:
>
> [eo at Haystack-S3] /interface wireless registration-table> /tool
> bandwidth-test 44.24.241.78 direction=receive
> status: running
> duration: 30s
> rx-current: 21.3Mbps
> rx-10-second-average: 19.4Mbps
> rx-total-average: 18.8Mbps
> lost-packets: 1720
> random-data: no
> direction: receive
> rx-size: 1500
>
> [eo at Haystack-S3] /interface wireless registration-table> /tool
> bandwidth-test 44.24.241.78 direction=transmit
> status: running
> duration: 30s
> tx-current: 25.2Mbps
> tx-10-second-average: 26.7Mbps
> tx-total-average: 24.4Mbps
> random-data: no
> direction: transmit
> tx-size: 1500
>
> But of course that's on S3, so he'll be sharing time domain with
> NR3O's uplink, so cut that speed at least in 1/2, and probably more
> when the -81dBm conditions return. Then we have a link to QueenAnne:
>
> [eo at QueenAnne.Haystack] > /tool bandwidth-test 44.24.242.39
> direction=receive
> status: running
> duration: 30s
> rx-current: 113.9Mbps
> rx-10-second-average: 124.3Mbps
> rx-total-average: 104.6Mbps
> lost-packets: 6234
> random-data: no
> direction: receive
> rx-size: 1500
>
> [eo at QueenAnne.Haystack] > /tool bandwidth-test 44.24.242.39
> direction=transmit
> status: running
> duration: 30s
> tx-current: 76.2Mbps
> tx-10-second-average: 76.7Mbps
> tx-total-average: 58.4Mbps
> random-data: no
> direction: transmit
> tx-size: 1500
>
> Which as you can see is pretty sweet, but it's then choked by the
> QueenAnne-Westin link when it comes to reaching the Internet:
>
> [eo at QueenAnne.Seattle] > /tool bandwidth-test 44.24.242.35
> direction=receive
> status: running
> duration: 30s
> rx-current: 3.2Mbps
> rx-10-second-average: 2.9Mbps
> rx-total-average: 2.9Mbps
> lost-packets: 28
> random-data: no
> direction: receive
> rx-size: 1500
>
> [eo at QueenAnne.Seattle] > /tool bandwidth-test 44.24.242.35
> direction=transmit
> status: running
> duration: 30s
> tx-current: 4.5Mbps
> tx-10-second-average: 4.3Mbps
> tx-total-average: 4.3Mbps
> random-data: no
> direction: transmit
> tx-size: 1500
>
> The Victoria link doesn't have direct peering to the Internet, it VPNs
> back to the Westin. Last time I measured, it does 15Mbit one way and
> 30Mbit the other way, and I don't know what the limits are on the
> cable modem VPN.
>
> So no, I would not consider Haystack as having good connectivity. We
> need to get Victoria onto a different cell site (yet to be
> constructed) so we can point the dish back at SnoDEM. Having the dish
> pointed back @ SnoDEM will also allow K7NVH's uplink to be collinear
> and get out of the -81dBm muck. We also need to fix QueenAnne's
> connectivity. This might have some chance of success as we get more
> access to the Capitol Park site. We could either up CP's connectivity
> with CA and Baldi, and route via Tukwila, or we could connect CP
> directly to Haystack and route via Tukwila. QueenAnne may also be able
> to link with Gold.
>
> So if you're seeing slow connectivity, on the order of 5Mbit or so,
> now you know why.
>
> PS: I disabled ch1 on Haystack.SnoDEM and the slow Haystack->SnoDEM
> transfer speeds didn't stop. Now I suspect it's the brand new 5GHz
> WiFi that got installed all over that building. Need to do more
> research and maybe find a better frequency.
>
> --Bart
>
>
>
> On 3/13/2017 11:27 PM, Rob Salsgiver wrote:
>>
>> Cool – thanks. Once again it seems my mind is the item suffering the
>> outages these days….
>>
>> Sigh
>>
>> Cheers,
>>
>> Rob
>>
>> *From:*PSDR [mailto:psdr-bounces at hamwan.org] *On Behalf Of *Nigel
>> Vander Houwen
>> *Sent:* Monday, March 13, 2017 11:21 PM
>> *To:* Puget Sound Data Ring
>> *Subject:* Re: [HamWAN PSDR] Haystack Site Outage
>>
>> Haystack is pretty well connected as well, it has links to Vancouver,
>> Paine, and Queen Anne, as well as Bart and myself are uplink nodes
>> connected to Haystack.
>>
>> Nigel
>>
>> On Mar 13, 2017, at 23:19, Rob Salsgiver <rob at nr3o.com
>> <mailto:rob at nr3o.com>> wrote:
>>
>> I was thinking more about the strength/stability of Haystack.
>> When I was connecting up the hospital in Monroe I thought I
>> remembered being told that Haystack was marginal due to the split
>> from Paine and one of the other links being interfered with
>> (Amazon?) – I might be mixing my site memories. If it’s marginal
>> I would like to see Haystack solidified (if possible) as I have
>> Evergreen on there and it’s my #1 option for Everett Clinic.
>>
>> Cheers,
>>
>> Rob
>>
>> *From:*PSDR [mailto:psdr-bounces at hamwan.org]*On Behalf Of*Nigel
>> Vander Houwen
>> *Sent:*Monday, March 13, 2017 10:48 PM
>> *To:*Puget Sound Data Ring
>> *Subject:*Re: [HamWAN PSDR] Haystack Site Outage
>>
>> Rob,
>>
>> That’s correct, the dish at haystack was turned a few degrees
>> farther north to better connect Victoria. However, the link to
>> Paine remains up, and has fairly good signal strength. Adding in
>> the East Tiger<->Paine and Gold<->Paine links, the site is pretty
>> well connected. Adding another dish to gain a few dB on an
>> already decent signal isn’t currently on the plans.
>>
>> Nigel
>>
>> On Mar 13, 2017, at 14:21, Rob Salsgiver <rob at nr3o.com
>> <mailto:rob at nr3o.com>> wrote:
>>
>> Nigel,
>>
>> If memory serves me correctly I think we swung the Paine dish
>> to cover the new Victoria site last year. Are there any
>> plans to install a separate dish to optimize both links once
>> the weather and access improve?
>>
>> Cheers,
>>
>> Rob / NR3O
>>
>> *From:*PSDR [mailto:psdr-bounces at hamwan.org]*On Behalf
>> Of*Nigel Vander Houwen
>> *Sent:*Sunday, March 12, 2017 11:24 PM
>> *To:*Puget Sound Data Ring <psdr at hamwan.org
>> <mailto:psdr at hamwan.org>>
>> *Subject:*Re: [HamWAN PSDR] Haystack Site Outage
>>
>> Evening Everybody,
>>
>> Today Bart and the site owner took another run at getting up
>> to haystack and managed to fight their way up to the site.
>> They managed to get a lot done, and Haystack is in much
>> better shape.
>>
>> * They found some misbehaving equipment that had a huge
>> draw (>400W continuous) on the power system up there,
>> overwhelming the solar generation capacity, draining the
>> batteries, and causing the generator to run more than it
>> should have been needed, and then run out of fuel.
>> * They brought up 40 gallons of fuel that was brought
>> partway up the mountain on the last trip.
>> * Bart replaced the Haystack-S3 modem to see if it would
>> fix an issue we’re seeing with non-symmetric signal
>> strengths. It didn’t resolve the issue, so we’ll revisit
>> that issue again later.
>> * Replaced the RJ45 connector for the Haystack-S3 modem
>> that may have been the cause for the earlier Haystack-S3
>> outage.
>>
>> All together, with the misbehaving loads removed from the
>> system, and the generator refueled to get the batteries
>> recharged so it’s not all on solar to bring things back up,
>> things are looking much better.
>>
>> We’ve powered back online all of the HamWAN equipment at the
>> site, so the sectors and backbone links are all up and
>> operating normally.
>>
>> I’d like to thank again the folks who made the previous
>> attempt and got the fuel most of the way up the mountain, as
>> well as bart and the site owner for making the run today to
>> finish the job. All the efforts are appreciated, and are
>> paying off in the site being back online.
>>
>> Nigel
>>
>> On Mar 4, 2017, at 23:41, Nigel Vander Houwen
>> <nigel at nigelvh.com <mailto:nigel at nigelvh.com>> wrote:
>>
>> Hello All,
>>
>> Our site at Haystack has been having some trouble with
>> the solar charging setup, and has not been able to get
>> much energy with the limited solar that remains working
>> onsite, as such, the site owner has been intermittently
>> running the generator to keep the batteries charged.
>> Unfortunately, as of Friday the generator is out of fuel.
>>
>> Bart, Bruce, and one or two others (I didn’t catch the
>> names), made a heroic effort today to get onsite with
>> snowmobiles and a snow cat to refuel the tanks, and
>> repair the solar charging problems, but were stymied by
>> snow conditions, and were not able to reach the site today.
>>
>> Considering the limited solar functioning, and the clear
>> weather today, we had a pretty good day, but the forecast
>> doesn’t look promising for keeping that up.
>>
>> I’ve shut down our sectors at Haystack to try and
>> conserve some power, so if you normally connect to
>> Haystack, or were looking to try, please be aware that
>> for the moment, the Haystack sectors are offline.
>>
>> We’ll let you know when we’re able to get the sectors
>> back online. Hopefully soon we’ll be able to get more
>> fuel up there, and the extra solar capacity repaired.
>>
>> Thanks again for the efforts of the volunteers that tried
>> to reach the site today. It was a real slog, and the
>> efforts are appreciated.
>>
>> Nigel
>> _______________________________________________
>> PSDR mailing list
>> PSDR at hamwan.org <mailto:PSDR at hamwan.org>
>> http://mail.hamwan.net/mailman/listinfo/psdr
>>
>> _______________________________________________
>> PSDR mailing list
>> PSDR at hamwan.org <mailto:PSDR at hamwan.org>
>> http://mail.hamwan.net/mailman/listinfo/psdr
>>
>> _______________________________________________
>> PSDR mailing list
>> PSDR at hamwan.org <mailto:PSDR at hamwan.org>
>> http://mail.hamwan.net/mailman/listinfo/psdr
>>
>>
>>
>> _______________________________________________
>> PSDR mailing list
>> PSDR at hamwan.org
>> http://mail.hamwan.net/mailman/listinfo/psdr
>
>
>
> _______________________________________________
> PSDR mailing list
> PSDR at hamwan.org
> http://mail.hamwan.net/mailman/listinfo/psdr
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://mail.hamwan.net/pipermail/psdr/attachments/20170314/350d6510/attachment-0001.html>
More information about the PSDR
mailing list