My Rambling Thoughts

Badge reels in all shades of blue

After the string on my badge reel broke (for the nth time; they are cheaply made and wear out every few months), and the Admin said she was out-of-stock for the foreseeable future, I thought, why not get my own?

One of my minor pet peeve is that the badge reel is not in the company color. Adding a logo is expensive, but surely get the color correct at least?

So I bought a bunch from TaoBao. Badge reel is 易拉扣. This is the key to finding them. They are really cheap.

The leftmost badge is supposed to be cyan, but it is bluish cyan instead. The second badge is a shade lighter than its picture. The third and fourth badges are grouped as dark blue, but it turns out to have two distinct shades (blue and dark blue).

The bottom left badge is the closest to company blue, but I like the transparent ones better.

Perhaps I bought from too cheap a seller. Many of the badges are scratched... but they still look okay from afar.

10 years of Pandemic

Pandemic: 10th Anniversary Edition (2018) [MSRP US$100]

Announced by Z-Man Games in July. Buy from them and get a limited edition with painted miniatures. Other retailers are listing it for US$85.

Will I get it? Most probably not. It is prettier, but the normal edition holds its own, especially if you take the price into account: US$85 vs US$35. And this is not compatible with the normal expansions. (Expansions will be released in this style if it does well.)

I already have both the first and second editions. :-O The second edition is already pretty nice. Plus, I have pimped the set with custom player pawns (US$15) and virus tokens (US$46).

This situation is similar to Power Grid Deluxe, US$52 vs US$34. Except I did buy it, and I always have second thoughts. The maps, resources and rules are changed — I prefer the original — and it is arguably uglier than normal PG. :lol:

In the past, I thought a 10th anniversary edition would be an automatic buy. It turns out not.

HDD SMART data 2018

DriveSizePower On HoursPower CyclesLoad Cycles
Primary2 TB55,757621771,881
data22 TB21,949904,458
Transporter (ext)1 TB7,5024323,573
OfflineStore (ext)2 TB3514412,623
BackupL2 (ext)3 TB2372641,161
infplus16 TB1,98424256
omega2*6 TB45622

*omega2 is an internal drive, but I use it with a trayless hot-swap mobile rack.

Nov 2016:

DriveSizePower On HoursPower CyclesLoad Cycles
Primary2 TB40,303593771,866
data22 TB9,487612,927
Transporter (ext)1 TB6,5493953,250
OfflineStore (ext)2 TB3434242,487
BackupL2 (ext)3 TB221237953
infplus1*6 TB38914118

*infplus1 is an internal drive, but I use it with a trayless hot-swap mobile rack.

April 2014:

DriveSizePower On HoursPower CyclesLoad Cycles
Primary2 TB17,61249771,793
data22 TB1675291
Transporter1 TB91790762
OfflineStore2 TB2181971,834
BackupL23 TB12669288

Note that I have named my drives, with the exception of the primary drive. :lol:

Scanning for bad blocks

In June, I downloaded a multi-part review of The Last Jedi, appropriately named A Critique of Star Wars - The Last Jedi [MauLer]. A few days later, I found that it could not be played back.

I immediately knew what it meant: bad sectors. However, I did not think the HDD developed new ones. Rather, these were missed in the previous scan — I only did one scan the last time. Now thinking back, that was very optimistic.

(I did not delete the file, but renamed it to .bad so that the sectors were not freed up.)

I kept putting off a full disk scan (even though the server would remain up) until now. Check file system for bad blocks (must be unmounted):

sudo e2fsck -k -c /dev/sdxn

-k preserves the existing list of bad blocks. Unfortunately, I overlooked it, so I lost the original list of bad blocks.

The first time I ran it, it took 2+ days and marked 1,233 bad blocks. The second time I ran it (with -k), it marked 1,272 bad blocks (+39). It took ~2 days. I think the existing bad blocks are skipped. The third time I ran it, it marked 1,275 bad blocks (+3).

Finally, I ran it with a second -c (non-destructive read-write test):

sudo e2fsck -k -c -c /dev/sdxn

1,277 bad blocks (+2).

Moral of the story: one scan is insufficient.

Interestingly, SMART data from 7/2017 shows 1,265 bad blocks. SMART data now shows 1,286 bad blocks.

I found an old list of bad blocks from 8/2013 and there were 2,188 entries. I wonder where the rest of the bad blocks went? Were they remapped by the drive?

The lastest Pi

One year after Stretch was released, I finally upgraded to it from Jessie. I put it off because the recommended way was to install it on a new SD card.

There is an official guide to upgrade in-place, but it is discouraged. It is really simple:

In /etc/apt/sources.list and /etc/apt/sources.list.d/raspi.list, rename jessie to stretch.

Then, run these:

sudo apt-get update
sudo apt-get -y dist-upgrade

(This is apparently the Debian way.)

It took over a day to download all the packages... the server is really slow.

The reason this approach is discouraged is that if things go wrong, you need to be reasonably comfortable issuing "arcane" Linux commands to make it work again. Luckily for me, I did not make much changes and all it did was ask me to choose between my modified config files or the new ones. I always choose the new one and re-apply my modifications after the installation, unless my changes are the only changes.

I have upgraded my Ubuntu in-place from 10.04 to 12.04 to 14.04 to 16.04, mostly problem-free. 18.04.01 is almost available, but I'll wait another three months. :-P

One thing I've wanted to do is to change to 64-bit version — 32-bit OS is passe these days. Almost everyone says a reinstallation is needed, that's why I've been putting it off. But I just read a post from someone who had done it — successfully. Doable, but scary. I need to think about it. :lol:

In-place upgrade is "fun", but the easiest and cleanest is to install 64-bit 18.04.01. I don't have that many customizations anyway.