I post­ed yes­ter­day that I liked Arch but I called it "not too good". So, Mark Kretschmann post­ed a com­ment ask­ing what I did­n't like.

It's not too much, but here it goes:

1. The up­­­grades some­­times are a bit painful (switch­ing to udev was a bit hard).

2. The pol­i­­cy of delet­ing the pack­­age doc­u­­men­­ta­­tion is evil. Re­al­­ly.

3. The start­up sys­tem is too sim­­plis­tic. No de­­fault or­der of start­up scripts means some­­times it takes tri­al and er­ror to fig­ure out what goes first. Hal or dbus? hwd?

4. The pack­­age se­lec­­tion (with­­out un­­sup­­port­ed) is some­what skimpy (no per­l-net-server? no per­l-ht­m­l-tem­­plate?) but that's prob­a­bly my POV be­­cause I am a bit server-ori­en­t­ed.

5. Some ba­sic pack­­ages make scary as­­sump­­tion­s. For ex­am­­ple, if you have a us­er with UID 89 when you in­­stall mysql server, weird things may hap­pen. Same for UID 40 and named.

On the oth­er hand, the good side (at least for an am­a­teur like me) is a bazil­lion times big­ger.

I have been on Arch for a while now.

So far, not bad. Not too good, ei­ther, but it is a fun dis­tro.

I am think­ing of us­ing it as a base for my ev­er-­va­porous per­son­al small serv­er dis­tro.

You can see any progress I make by search­ing for my pack­ages in AUR (search for sub­mit­ter ralsi­na). AUR is re­al­ly a nice thing to have for a hob­by­ist. You put your stuff there, peo­ple can find it, they can even try it, com­ment on it, vote on it... re­al­ly a very nice com­mu­ni­ty site for a very hob­by­ist dis­tro.

At least in the sense that they are re­al­ly full of peo­ple with bad in­ten­tion­s...

• The red stripe are con­nec­­tions blocked be­­cause the sender is list­ed in spam­­cop.

• The blue stripe is con­nec­­tions blocked be­­cause the sender is in NJABL.

• The dark­­er green stripe is con­nec­­tions that are rate-lim­it­ed be­­cause they look like a DOS or ad­­dress spam (client con­nec­t­ing too quick­­­ly)

The top green line is SMTP con­nec­tion­s.

The thin white stripe is re­al con­nec­tions that we al­low to send email.

Of course when you look at those...

• 42% are spam (yes, the fil­ter is quite ef­­fi­­cien­t).

So, about 1 in 5 con­nec­tion at­tempts ac­tu­al­ly in­tends to send a re­al, good, valid email.

I am amazed this email thinga­ma­jig still work­s.

Tra­di­tion­al­ly, you graphed your sys­tem's sta­tus us­ing MRTG but nowa­days, there are much nicer tool­s, and I like Munin.

For MRT­G+q­mail there are many script­s, but I could­n't find one I liked for Munin.

Since munin is mod­u­lar, it's easy to fix, of course.

First, get the ex­cel­lent qm­rtg which gives you great, quick, awe­some mul­ti­log-crunch­ing tool­s.

Then, check the cod­s, check the munin plug­in doc­s, and it's pret­ty much a 20-­line thingie. In shel­l.

I would post it, but my blog­ging tool hates shell code :-)

And you get your graph­s. I am prob­a­bly go­ing to write a col­lec­tion of these and pub­lish it some­where.

No, do not use gdb­m. Prob­a­bly not bdb ei­ther.

TDB al­lows for con­cur­rent writes. So, if you want to store some da­ta in a file in your ap­pli­ca­tion/­tool/­com­mand/what­ev­er, and there is a chance that you need con­cur­rent writes... TDB is re­al­ly the on­ly rea­son­able choice I know.

You can even have dis­trib­uted TDB­s.

I found out about TDB while writ­ing a per­sis­tant cache for a lit­tle tool (a qmail plug­in). Since by its na­ture it would have con­cur­rent writes all the time, I was aim­ing for some­thing like mem­cached, or its disk-based ver­sion... but TDB solves the prob­lem nice­ly and ef­fort­less­ly.

So, if you need a flat DB (a.k.a anony­mous blobs with a sin­gle in­dex, a.k.a a hash in a file)...