Microsoft sends me a periodic email (TechNet Flash Newsletter) listing the news related to their product and ecosystem that I read in a sporadic way, but a few weeks ago thanks to the Christmas holidays I had a bit of extra time and read through one containing an invite to a challenge on hackster.io.
Joining the community was quick and straightforward.
After a few days I put in an idea for the challenge pre-contest and earlier today I found out that it was selected and I should get the Genuino MKR1000 to make it a reality.
Tools are installed on my W10 phone and notebook ready to consume my week-end spare time for the next 46 days.
I believe this marketing initiative is a very smart one giving good visibility of MS tools in the IoT space to the people who should really care (developers and tinkerers) and can create the tools, applications and devices that will feed the Azure infrastructure with major volumes of data in the coming years.
I posted about my early experiences with this technology about 8 months ago here and here.
At the time of these posts the apartment was perfectly empty except for the adapter and the notebook I was using.
Now, after a full renovation including the electrical infrastructure, it is a real home with all the associated devices and appliances connected and operating.
Another important change is that, due to the limits of the electrical tubing connecting upper and lower level, I had to connect the two levels using the same technology and now I have in place 4 adapters from the same manufacturer.
What I read right now in the monitor from the upper level is:
110 Mbit/s to the lower level
70 Mbit/s to the garage
60 Mbit/s to the underground room.
To the garage what I get at the application level moving data to the garage is about 3MB/s (a bit shy of 30Mbit/s) when using a backup program targeting a share on the LS220D and about half of this when using directly a samba share on the same device to move files with windows explorer.
The number reported are in the lowest range of the day and they can easily get 30% better than this depending on the amount of electrical noise tha is injected on the line from the other apartments on my building.
Marketing proves to be even more distant from the reality than I already complained about. But it is still better than WiFi in my environment.
I’m experiencing some significant brownouts during the day with the worst quality during the evening and at lunch time.
This might also be a factor, but I have opened a complaint and in a couple of months should be addressed by my electricity distribution company.
I am doing a backup using disk2vhd from former sysinternals (now Microsoft) Mark Russinovich and I’ve found something puzzling.
The source is a W7 ultimate machine, the destination is a SMB share on a W7Pro connected to the same 1Gb/s Ethernet switch.
The two systems are not doing any other significant network activity but I see resource monitor reporting very different network use on the two sides of the transfer.
I am monitoring the destination machine from the source machine using teamviewer (great tool free for personal use) and here is the picture of the two paired resource monitors:
The source machine (on the right side) reports a network use that is significantly lower than the destination machine (on the left side).
The source machine also shows that the largest transfer alone (the backup) has a higher throughput than the total reported.
I was thinking the discrepancy might be related to a time drift in the reporting, but the graphs show that is not the case: the destination system constantly reports a higher network use.
Does anyone know if it is a known bug in W7 and how to fix it?
Two days ago I did a new bios update on the notebook.
The process worked fine as usual and, again as usual, did not fix or improve the issue with the fan noise.
After 36 months with it I have to bear it for only 12 more months until the notebook is due to refresh.
A positive note about the 6430U: it does no longer trigger the security scanner in the Ben Gurion airport. Whatever the chemical that was there it is now completely evaporated.
A month after my first installation I gained additional insight about my electrical layout.
It turns out that my apartment and my basement/underground room are not directly wired: I can turn off the meter of the apartment while I still have power in the garage and underground room.
This means that most likely only ground and neutral are shared.
In the best case the phase is hared too but has multiple elements in the middle: at least two meters.
Given the circumstances the 60Mbit/s are not too bad.
Changing the plug type in the basement fitting a new round “German” type and removing the adapter I was using earlier did not change the signal quality.
On the bad side: I tried the firmware upgrade of the adapters and now the powerline utilities can’t talk with the adapters anymore.
Yet the adapters talk to each other: I’m going to keep this setup until the endo of warranty approaches or I get a proper cat 5e cable to connect the two places.
In a couple of months I will move into a new apartment and with this comes a new underground room and a new garage that I want to connect to the apartment.
Wifi is not really an option in this environment due to the relative positioning of the spaces and I decided to give another try at homeplug.
In the past I had a base HomeplugAV and the performance was a disappointment.
Better than wi-fi in the specific location that I needed to reach (where wifi signal was nonexistent) and useable for light internet navigation, but not an acceptable transmission layer to do a backup.
This time I went for the latest and greatest incarnation of the standard to get another disappointment.
Hope: 350Mbit/s reported by the monitoring tool in the garage and 186MBit/s in the underground room.
Reality: 60-70Mbit/s in samba file transfer
It is much faster than the earlier HomeplugAV, but the marketing-reality gap is still embarrassing.
The product is clearly geared toward IT-illiterates: the monitoring tool is extremely limited.
Sometime the tool goes nuts completely: it claimed that the local adapter was not connected while at the same time a file transfer was happily ongoing.
The price for the TP Link PA8030P kit is not popular at the moment, but you get 3 gigabit ethernet ports that in many cases removes the need for an additional switch and a leading edge technology.
Should I find a way to bring an ethernet cable from the apartment I’ll quickly dump (again) homeplug, but until then it is better than nothing, better than wi-fi in the specific location and better than the older versions of the standard.
Should you have to use your electrical wiring for data then go for AV2 with MIMO: the saving obtained by using the older versions quickly vanishes once you start experiencing the even lower performance associated with the earlier version.
I’ve reported earlier that my DS411Slim started to have critical overheating problems and that I did replace the 3 years old system fan without success.
In this second fix attempt I purchased a thermal adhesive from Akasa and using a filer I adapted a heat sink that I had hanging around so that it would fit with the bolts that keep the heat spreader attached to the electronic board.
High hopes as before, but the file compression test failed again. The interface is working and the heat sink gets very hot.
Unfortunately it seems that the more heat I remove from the CPU the more gets generated.
Indexing videos generated the very same behavior.
I’m running out of ideas and I’m starting to challenge the idea of getting a DS414Slim to make a fail-resistant setup.
I see no point in investing in another Synology as the first one is constantly failing.
It is too bad because the management software is very nice and powerful.
Anyone that has an idea on what to try next is welcome to comment to this post.
Below are a few snapshot of the hack.
Factory fit heat spreader with connecting bolts that require heat sink adaptation.
1st corner of the heat sink filed
2nd corner filed
Interference check with the filed heat sink before bonding
Side view of the bonded heat sink
Side view with the 3 drives in place before the insertion of the empty disk frame.
Many thanks go to Mark on DIY forums that pointed out the issue with the switch of the charger here
0.25€, a bit of patience (and a bit for a triangular screw head too) and the battery charger is back to work.
I’ve reported a problem with my nas a few weeks ago (here) and my plan to get it back to productive work in a reliable way.
As part of this plan I did order a OEM replacement fan: searching an equivalent model with the same small connector proved to be a waste of time.
After a few weeks it finally arrived and yesterday I picked it up at the shop and replaced it in the NAS.
The old fan had some substantial play in the axial direction and at low-speed failed to spin up, for this reason I had high hopes with the replacement.
It was a quick job requiring only 15 minutes and then I was ready to use the NAS.
Mirror rebuild worked fine, compressing a file with the PC CPU with source and destination on the NAS worked fine.
Active disks were at 43C and the CPU at 48C during the rebuild; a few minutes after it was completed the disks were down to 35C and the CPU at 43C.
With 19C of ambient temperature this is not an extraordinary achievement, but are manageable temperatures.
I felt confident, even if not overly confident, that the fix was going to work as desired.
To be sure that this was really the case and that the overheating was not going to hit me unexpected in the future I did a heat torture test by compressing a 140GB file using the NAS CPU instead of the PC CPU.
This time unfortunately the NAS did behave as with the older fan: after some time it died on me.
The latest reported CPU temperature in the web interface was 54C and it survived long enough to put a warning in the web interface telling me that it could not compress the file.
The NAS couldn’t be reached anymore (no web interface, no ping, no Synology assistant) and couldn’t even be restarted using the power button.
Next step will be to mount a heat sink on the current nonremovable stock heatspreader.
I have already found a suitable heat sink, now the challenge is to find the needed adhesive thermal pad.
I will lose the ability to use the 4th disk slot, but it is acceptable because future space expansions will likely come via a DS414slim for additional redundancy and performance.
And I still have the LS220D that can maintain some of the lower value data.
At the end of 2014 my Synology DS411slim decided to abandon me and a couple of weeks ago I have purchased this nas to store the data in the Synology until I could put some more thinking into an appropriate fix/hack/replacement.
I expected that the LS220 would be no match for the higher end brands in the SOHO area, but still I couldn’t help the first thought: cheap plastic.
Once it is filled with the two disks it feels a bit better thanks to the extra weight and reduced flex that they warrant to the device.
The setup was fairly smooth even if the quick startup guide could be better and the UI was all but exciting.
The same feeling as before: cheap.
In the end all I need in the short run is a SMB share to copy as much data as possible from the DS411Slim while it works before overheating.
For this basic use I consider the device as a good fit for a very convenient price, but then comes in the performance in my environment.
The best sustained writing rate I could get so far is 28MB/s on a cabled 1Gbit link while using disk2vhd to backup my notebook (98GB in about 57 minutes).
Compared to the 37-40MB/s that I was getting on the 4 years old DS411Slim (see this old post about encrypted vs. unencrypted writing speed I was experiencing) this is not a good number for this workload and makes the LS220D less suitable for one of my key use cases.
To conclude: if you have a pair of spare 3.5″ drives in a drawer the LS220D is an expensive way (I paid mine 69.99€) to put them to work for the benefit of the entire house.
If you are planning to purchase the drives for the purpose of putting together a NAS you should really shop for a higher class NAS than the LS220D.