Baldi Deployment Report: Part Deux
Better late than never! We visited Baldi instead of Rattlesnake last Saturday. Both sites were in rain, and Baldi had the easier set of tasks, so we opted for the safer option. Here's the work report: 1) Baldi has had a server so old for so many years that it's RAID drivers are no longer supported by HPE for management within ESXi. It's also had very little RAM (8GB) and drive space (RAID1 of 73GB drives). I took one of my spare servers to serve as a new hypervisor up there. It has 48GB RAM and 4x250GB SSD in RAID10 (500GB space). After install, I discovered iLO had somehow lost all the credentials I saved into it. It took all of Sunday to setup a special PXE environment + OS to finally gain control of iLO from the server side (not network side). Everything here is finally working as it should, and VMs will be migrated soon so we can decommission the old server. 2) The switch in "Building 2" at Baldi was full! We managed to install a new 24-port gigabit switch, pictured here: https://photos.app.goo.gl/8a98D44dDxkeqJn37 This switch is important as it links "Building 1", and "Building 3". We also discovered the UPS powering all the repeaters and the network gear was overloaded. It would beep when things were transmitting. We re-arranged the load so that it's only backing up the network gear now, and the repeaters will need their own larger UPS. 3) We installed the shield on the CampMurray facing dish. Over the last week I noticed the VPol signal on the link was really low (-94dBm) whereas the HPol was just fine (-65dBm). Prior to shield install on the dish, I inspected the connectors for any water ingress. I did not find any. However, I did find what looked like crusty build-up on the center pin of the VPol RP-SMA-to-N adapter. This adapter was re-used from the old 3ft dish. That dish had a split-open HPol connector (likely caused by ice expansion), which I suspect allowed water to pool in the feedhorn and probably drip down to this adapter, causing it to build-up whatever this grey substance is. We did our best to clean up the connector with the tools at hand. After re-installing and re-aligning the dish, VPol is now reading -84dBm instead of -94dBm. While this is a 10dB improvement, there's still clearly a problem. We need to return with a new adapter, and a method to properly clean the female N connector on the antenna to remove any residue which might still be there. Possibly re-compress the pin if it was split open too much by the residue. Another thing which might be affecting VPol is misalignment of the corresponding dish at CampMurray. That dish was never aligned with Baldi for lack of ever hearing a signal. If it's severely off-axis in the horizontal, the VPol transmission might be getting shunted by its shield (conductive vertically when azimuth is off). Both things need to be addressed before this link can be considered done. 4) We had serious Ethernet link stability problems with the new Rattlesnake.Baldi dish. Its cable is routed just a few feet from an FM broadcast antenna, and several repeater antennas. We installed a 30m (100ft) run of multimode fiber to connect the Rattlenake.Baldi modem to the switch in Building 3. The link came up and has been rock solid. I also suspected there might have been water in the original RJ45 leading to these Ethernet link outages. This was inspected and the connector was found to be dry, so there's no worry of corrosion due to PoE voltage. While the cause of the CAT5 link failure is not known, I think we can narrow it down to either RFI or a bad crimp job on the tower-side connector. The shack-end connector was re-done twice with no change in link failures. Anyway, this modem is now happy with the fiber and has great signal to Rattlesnake (-55dBm)! Now let's hope for some nice weather for the next visit to Rattlesnake, which requires work at about the 150ft level of a tower. --Bart
Thanks Bart! On Tue, Jun 16, 2020 at 12:21 PM Bart Kus <me@bartk.us> wrote:
Better late than never!
We visited Baldi instead of Rattlesnake last Saturday. Both sites were in rain, and Baldi had the easier set of tasks, so we opted for the safer option. Here's the work report:
1) Baldi has had a server so old for so many years that it's RAID drivers are no longer supported by HPE for management within ESXi. It's also had very little RAM (8GB) and drive space (RAID1 of 73GB drives). I took one of my spare servers to serve as a new hypervisor up there. It has 48GB RAM and 4x250GB SSD in RAID10 (500GB space). After install, I discovered iLO had somehow lost all the credentials I saved into it. It took all of Sunday to setup a special PXE environment + OS to finally gain control of iLO from the server side (not network side). Everything here is finally working as it should, and VMs will be migrated soon so we can decommission the old server.
2) The switch in "Building 2" at Baldi was full! We managed to install a new 24-port gigabit switch, pictured here: https://photos.app.goo.gl/8a98D44dDxkeqJn37 This switch is important as it links "Building 1", and "Building 3". We also discovered the UPS powering all the repeaters and the network gear was overloaded. It would beep when things were transmitting. We re-arranged the load so that it's only backing up the network gear now, and the repeaters will need their own larger UPS.
3) We installed the shield on the CampMurray facing dish. Over the last week I noticed the VPol signal on the link was really low (-94dBm) whereas the HPol was just fine (-65dBm). Prior to shield install on the dish, I inspected the connectors for any water ingress. I did not find any. However, I did find what looked like crusty build-up on the center pin of the VPol RP-SMA-to-N adapter. This adapter was re-used from the old 3ft dish. That dish had a split-open HPol connector (likely caused by ice expansion), which I suspect allowed water to pool in the feedhorn and probably drip down to this adapter, causing it to build-up whatever this grey substance is. We did our best to clean up the connector with the tools at hand. After re-installing and re-aligning the dish, VPol is now reading -84dBm instead of -94dBm. While this is a 10dB improvement, there's still clearly a problem. We need to return with a new adapter, and a method to properly clean the female N connector on the antenna to remove any residue which might still be there. Possibly re-compress the pin if it was split open too much by the residue.
Another thing which might be affecting VPol is misalignment of the corresponding dish at CampMurray. That dish was never aligned with Baldi for lack of ever hearing a signal. If it's severely off-axis in the horizontal, the VPol transmission might be getting shunted by its shield (conductive vertically when azimuth is off). Both things need to be addressed before this link can be considered done.
4) We had serious Ethernet link stability problems with the new Rattlesnake.Baldi dish. Its cable is routed just a few feet from an FM broadcast antenna, and several repeater antennas. We installed a 30m (100ft) run of multimode fiber to connect the Rattlenake.Baldi modem to the switch in Building 3. The link came up and has been rock solid. I also suspected there might have been water in the original RJ45 leading to these Ethernet link outages. This was inspected and the connector was found to be dry, so there's no worry of corrosion due to PoE voltage. While the cause of the CAT5 link failure is not known, I think we can narrow it down to either RFI or a bad crimp job on the tower-side connector. The shack-end connector was re-done twice with no change in link failures. Anyway, this modem is now happy with the fiber and has great signal to Rattlesnake (-55dBm)!
Now let's hope for some nice weather for the next visit to Rattlesnake, which requires work at about the 150ft level of a tower.
--Bart
_______________________________________________ PSDR mailing list PSDR@hamwan.org http://mail.hamwan.net/mailman/listinfo/psdr
participants (2)
-
Bart Kus -
Doug Kingston