APC UPS Smart-UPS 1400VA (SU1400RMNET)

3dd05c6b124fe0a42efb6194097a1-orig

An old unit but getting 12-month warranty for the battery and for the price of 130 € including an SNMP communications card.

These are also simple and easy to fix if they go wrong.

It is the same model as one that I have had on my eBay watchlist for a year, but haven’t bought it because of 200€+ price. This is locally available so no postage. Pure sinewave, which is the only one to really consider.

Expecting the new infrastructure take perhaps 420 watts, which would give maybe 20 minutes. More than enough to save everything and even wait for the power to come back.

Regulation

I have a halogen lamp hooked to the device and the light level is clearly oscillating. So either it isn’t pure sinewave, or it is doing poor job regulating. Could be old caps. But I don’t have tools to diagnose the problem so for now I just have to trust that the power is good. That’s a poor predicament but what can I do.

Charges at 75W so definitely not happy with that. At that rate it will take 10 hours, or so, to recharge. Perhaps this is the way all UPS work but that feels like poor performance. Perhaps it has something to do with battery design or longevity of the batteries, or the battery technology.

Installation succeeded

First of all I must say that whenever you can: get a dual PSU machines; there were no distractions to my main server when the UPS was plugged between the machine and the wall outlet, while another server with single power supply suffered from me not being even close fast enough in switching the power.

But this UPS isn’t for keeping the machines up and running for a long time, that is for sure. The batteries were completely drained after only maybe three minutes. So either the batteries aren’t the best or time went by really fast. But I will be doing complete test down to 1 bar of capacity once the batteries have recharged. Also seriously considering to get another one since the man said he has plenty of these. And the price is on the spot.

More about the current setup

So the way I have set things up is like this, looking from the wall outlet onwards:

  1. Ground fault circuit interrupter
  2. Power meter
  3. Overcurrent protector
  4. UPS
  5. Sensitive devices

So the idea is that if there is current leaking to wrong places, then the ground fault circuit interrupter will save lives and hopefully prevent fires, and when everything works correctly, the power meter will measure the power usage, the overcurrent protector will protect the UPS, and the UPS will protect the devices (secondary function) and provide them power if utility power fails.

The location of power meter is probably wrong because it is now receptive to overcurrent, and could in such situation cause fire. So it will be moved between the overcurrent protector and the rest of the system.

The power usage of this model, by the way, is about 40W idle. AFAIK it is line-interactive and not online so not sure where it puts all that power. 40W is quite a bit.

The noise is practically non-existent. Maybe if you have 100% silent setup, then the noise might be overpowering, but for anyone buying devices like these then no, not really a concern.

Related Post

New workstation for work?

So my superior has been pushing us to get upgrades on our machines and while I gave it a thought, I quickly abandoned the idea because you cannot get proper machine for “reasonable amount of money”, as he might have put it.

So I think I now have a plan and a machine which I want to have:

s-l1600

That is of course a used machine but it is Intel Xeon E5-2660 with 16 threads and 64GB of ECC DDR3. Then on top of that I figured I would throw 500GB SSD and two Western Digital Red 2TB for mass storage.

The memory thing is interesting because I have a need sometimes to go through huge masses of data, and with 64GB of memory, that whole set of data would fit in the memory. And the dataset would be stored on the SSD, for fast initial read into memory.

You could even go so far as to split 32GB off of the memory to make a ram disk and work 100% in-memory. And the Xeon with large caches would make the whole thing that much faster.

That is what I call a workstation. A proper professional setup.

SMART disk errors

I put this here so that I myself remember these numbers and if they increase.

cciss,2

Error counter log:
           Errors Corrected by           Total   Correction     Gigabytes    Total
               ECC          rereads/    errors   algorithm      processed    uncorrected
           fast | delayed   rewrites  corrected  invocations   [10^9 bytes]  errors
read:   66321705        0         0  66321705   66321705     813856.124           0
write:         0        0         0         0          0       4507.073           0
verify:      956        0         0       956        956         36.420           0

Non-medium error count:    93997

+7 days:

Error counter log:
           Errors Corrected by           Total   Correction     Gigabytes    Total
               ECC          rereads/    errors   algorithm      processed    uncorrected
           fast | delayed   rewrites  corrected  invocations   [10^9 bytes]  errors
read:   66327001        0         0  66327001   66327001     813922.785           0
write:         0        0         0         0          0       4509.358           0
verify:      956        0         0       956        956         36.420           0

Non-medium error count:    94047

I don’t know what the deal with this is but the number is way way larger than with any other drives.

cciss,3

Error counter log:
           Errors Corrected by           Total   Correction     Gigabytes    Total
               ECC          rereads/    errors   algorithm      processed    uncorrected
           fast | delayed   rewrites  corrected  invocations   [10^9 bytes]  errors
read:          0    75669         0         0          0     469053.111           0
write:         0       14         0         0          0       2340.964           0
verify:        0      976         0         0          0       1529.626           0

Non-medium error count:      156

+7 days:

Error counter log:
           Errors Corrected by           Total   Correction     Gigabytes    Total
               ECC          rereads/    errors   algorithm      processed    uncorrected
           fast | delayed   rewrites  corrected  invocations   [10^9 bytes]  errors
read:          0    75671         0         0          0     469106.610           0
write:         0       14         0         0          0       2343.320           0
verify:        0      976         0         0          0       1529.626           0

Non-medium error count:      156

cciss,4

Error counter log:
           Errors Corrected by           Total   Correction     Gigabytes    Total
               ECC          rereads/    errors   algorithm      processed    uncorrected
           fast | delayed   rewrites  corrected  invocations   [10^9 bytes]  errors
read:          0     3457         0         0          0     194024.310           0
write:         0       76         0         0          0       3751.003           0
verify:        0        0         0         0          0         36.420           0

Non-medium error count:      338

+7 days:

Error counter log:
           Errors Corrected by           Total   Correction     Gigabytes    Total
               ECC          rereads/    errors   algorithm      processed    uncorrected
           fast | delayed   rewrites  corrected  invocations   [10^9 bytes]  errors
read:          0     3459         0         0          0     194070.717           0
write:         0       76         0         0          0       3753.249           0
verify:        0        0         0         0          0         36.420           0

Non-medium error count:      338

cciss,5

Error counter log:
           Errors Corrected by           Total   Correction     Gigabytes    Total
               ECC          rereads/    errors   algorithm      processed    uncorrected
           fast | delayed   rewrites  corrected  invocations   [10^9 bytes]  errors
read:   54100861      237         0         0          0     336391.552           0
write:         0        0         0         0          0       1661.110           0

Non-medium error count:      141

+7 days:

Error counter log:
           Errors Corrected by           Total   Correction     Gigabytes    Total
               ECC          rereads/    errors   algorithm      processed    uncorrected
           fast | delayed   rewrites  corrected  invocations   [10^9 bytes]  errors
read:   54227591      237         0         0          0     338371.857           0
write:         0        0         0         0          0       1663.441           0

Non-medium error count:      141

 

Related Post

Gaining on infrastructure

Gaining more infrastructure in form of one of these for 145 €:

serveimage (4)

And one of these for 16 €:

s-l1600 (3)

So these are obviously SAS controller and an external LTO-4 tape drive. I finally spared to spend the money to get the drive. Because this one was cheap. That is 1700 € drive new in Finland. I hope it works. It was advertised to be in great condition. And so if it works and stays in good condition for my once a month two-tape usage, then it was an excellent bargain.

Because tapes just are so much better for backups. You put them in device, you write them, take them out, and move to safe location. Then you bring old tapes back, write them, and the cycle continuous.

The price is little bit higher than that of a hard drive, but hard drives don’t have any of the tape advantages.

If it works, then what I will do is I will probably archive ALL of the historical data to tape in two copies, at two different locations, and then I don’t need to worry about all that data no more. Because data has tendency to grow in a cancerous manner. And the more you move old data around, and the more you do and test; the more old data you create.

So 800GB tapes are more than enough to each store quite a bunch of tests. For 30 € a pop.

 

Related Post

Non-virtualized Windows??

I know it’s a crazy idea but on the other hand: why the hell not?

I earlier dismissed Windows for their licensing costs but I have managed to acquire some licenses, so now I have the possibility to keep using and learning more.

So perhaps I will put Windows Server 2012 R2 on physical hardware. It would let me get more out of it. Perhaps try virtualization and things like that.

It could be done with the spare machine (from previous post). One pair of disk would host backup Linux and the other pair would be Windows testing server and the rest for storage.

I think that’s a good plan. I might also get Windows Server 2008 R2 and Windows 10 while I am it.

Related Post

Backing up the old infrastructure

snap1475

It took some time to finally realize that the “Server name:” in “Connect to Server” was NAMEOFMACHINE\INSTANCENAME.

snap1476

Related Post

Giving up on Windows and VMWare

I will be abandoning my VMWare based plan and infrastructure due to the fact that VMWare and Windows based infrastructure is too complicated to maintain.

Simple things are overly complicated and while these two may serve well in a large complicated environment; for small special purpose systems they create too big of an overhead.

So the next setup will be based on Linux and small Windows network will be added as a companion for tasks which require Windows, and for Windows based testing.

Perhaps in the future I will spend more time with this type of setup, and at work I know I will, and I am happy for it; but for my personal setup I intend to stay with Linux for its simplicity.

Future setup will probably be something like this:

  1. 8 Core X5355, 32GB, 2x2TB for backups and storage
  2. 6 Core Opteron 2431, 32GB, 4x300GB + 4x146GB for virtualization

And then I will leave 8 Core 32GB Opteron 2356 as a backup.

I can get so much more out of Linux than I can out of VMWare and Windows.

It is sort of a sad decision to make but for the better. I would not want to give up on VMWare and Windows but the complixity and the difficulty to get simple things done is just too big of a pain. It is not worth the time and effort. Also from resource point of view; Linux is such a great tool when it comes to maximizing the potential of hardware, because the software is they key, and so many people have made so many great tools and contributions, that no other system can match what is available, and for free.

Also Windows licensing costs are prohibitively expensive. Even the most basic version of Windows 2012 R2 costs in excess of 600 € which for personal, hobbyist use is not going to happen that easily. It is a huge investment.

This two-server setup should let me run everything I need with minimum amount of electricity. My bill for the past 3 months of testing was 100 € a month which is also too big of a price to pay to learn Window and VMWare. But at least it was an intensive 3-4 months of studying which will continue, but not at the center of things. That role is for UNIX.

Also since I now discovered how qcow2 can be used to thinly privision virtual machines, all this put together gives me a great starting point towards even better design. I now have the possibility to save so much disk space that 600GB should be enough for all my virtualization.

Also what I have learned is that there are several different needs such as encrypted onlive content, encrypted offline content and encrypted hot online content, which can be hundreds of gigabytes. And while playing with fast SAS disks is fun and they are reliable pieces of hardware; they simply are way too small to handle it all.

So there are many things that must be taken into consideration when designing this. It will be extremely interesting to put it together better than ever before. So many mistakes were made in attempt to refactor the infrastructure that with that knowledge learned, the new infrastructure will be built on way more solid grounds.

New base image strategy

-r-----r--. 1 root root  26G Jul 31 15:21 2-windows2012r2
-r-----r--. 1 qemu qemu 194K Jul 31 15:37 2-windows2012r2-rev0
-rw-r-----. 1 qemu qemu  26M Jul 31 15:43 2-windows2012r2-rev0-vm1

The idea is that the 26GB file is the absolute base image on top of which multiple revisions can be used on top of which there can be multiple virtual machines.

This is because the base image will get old over time and new revisions may need to be created. But because these new revisions most likely will not take much space, it is cost effective to have that additional layer, versus always creating new complete 26GB base file.

26MB of difference was created after first boot.

Also learned from the history, the sources\sxs is not included within the base image so that all features are easy to install.

There is also performance increase because now every server references the same blocks on the file system and the underlying intelligent filesystem can better manage the blocks and keep them in cache.

Also theoretically the base image could be stored on highly compressed volume while the differential would be stored on non or slightly compressed volume. It would be balanced act between least amount of storage space used vs. fast access time to instance specific data.

This change will save me about 30GiB which is a lot considering the limited space, and the price of larger SAS disks.