IRC logs for #opendune on OFTC at 2009-12-03
⏴ go to previous day
00:55:16 *** DorpsGek has joined #openDune
00:55:16 *** ChanServ sets mode: +o DorpsGek
10:28:54 *** ChanServ sets mode: +v DorpsGek
10:28:54 *** ChanServ sets mode: -o TrueBrain
10:29:04 *** ChanServ sets mode: +v Xaroth
10:44:30 <TrueBrain> hmmm ... openttd server crashed last night
10:44:35 <TrueBrain> now one of hte raid devices did not recover
10:49:53 <TrueBrain> always nice to recover a raid device of 480GB on a hot system ...
10:49:57 <TrueBrain> it goes with 24MiB/s :p
10:50:04 <TrueBrain> @calc 480 * 1024 / 24 / 3600
10:50:04 <DorpsGek> TrueBrain: 5.68888888889
10:58:56 <Xaroth> we have a 1.8tb raid6 array that hickups every once in a blue moon
10:59:12 <Xaroth> but that usually does a rebuild within 3 hours
10:59:25 <Xaroth> while running live, ofcourse
11:03:22 <TrueBrain> says enough about the IO usage :p
11:12:44 <TrueBrain> well, that was a good waste of my morning
12:17:10 <Xaroth> TrueBrain: and the fact those servers have the best performing hard drives we could find :P
12:19:47 <TrueBrain> Xaroth: it is not the disks :)
12:20:02 <TrueBrain> the softraid limits the transmission speed of remirroring to allow good performance while doing so
12:20:24 <TrueBrain> in other words, if capacity is N, the current average transfer rate is M, it allow at most remirroring at N - M * 4
12:20:31 <TrueBrain> with a bottom of a few MB per second
12:21:03 <TrueBrain> this is a very simple and very efficient algorithm to avoid problems on production systems because of a remirroring processes
12:21:15 <TrueBrain> you can easily bypass it, but that will cost performance :)
12:21:26 <TrueBrain> not everything that is fast, is better ;)
12:31:40 <TrueBrain> of course; hardraid is not worth the money
12:32:06 <TrueBrain> hardraid btw still does such calculations and limit the rate of cloning
12:32:19 <TrueBrain> it took longer to be included in softraid then it was in hardraid ;)
12:33:05 <Xaroth> something tells me dih rebooted something :o
12:34:07 *** planetmaker has joined #openDune
12:34:13 *** planetmaker is now known as Guest1115
12:43:03 *** Guest1115 is now known as planetmaker
12:47:41 <TrueBrain> [UPDATE 07:20] A customer in Hall 2 who was replacing network equipment caused a loop which caused partial unreachability in Hall2.
12:47:56 <TrueBrain> LeaseWeb had a powerdip in one of their feeds, kicking few systems offline ...
12:48:01 <TrueBrain> under which ours :(
12:52:46 <TrueBrain> hmm .. reading more reports on EVO, it seems feed A was down for a few secs ...
13:03:21 <TrueBrain> the UPS went in static bypass for a few secs, which caused a power dip in a feed, which knocked down a good part of the whole DC
13:03:41 <TrueBrain> most shit rebooted instantly (like ours), but some systems had a hard time dealing with a power dip (mostly those who did not reboot)
13:03:55 <TrueBrain> it is known when you send a powerdip to a server, things like network goes terrible wrong ;)
13:05:51 <TrueBrain> even more fun to send a powerdip to a switch
13:05:56 <TrueBrain> you can lose like 50% of your ports
13:06:02 <TrueBrain> once took me a while to figure out what the fuck was wrong :p
13:06:21 <TrueBrain> Xaroth: you had no problems?
13:09:12 <Xaroth> where do you see those reports, TrueBrain ?
13:09:20 <TrueBrain> Xaroth: everywhere :)
13:09:26 <TrueBrain> EVO of course official says nothing went wrong
13:09:28 <TrueBrain> but they always do that
13:09:45 <TrueBrain> people who were there at around 01:53, saw the lights go out and shit :p
13:10:01 <TrueBrain> still security personal told them nothing was wrong .. yeah .. one of the DNS forwarders was offline
13:10:14 <TrueBrain> but okay, mostly H1 and H2 was affected
13:10:57 <TrueBrain> our company servers not ;) The core switch of our supplier did :p
13:16:14 <Xaroth> we're in H3 so we noticed nothing :P
13:48:10 <glx> TrueBrain: I know why Script_General_GetDistance is "general", it's because it was in 0EDB like other general scripts
15:13:50 <glx> hmm weird, sometimes 6344 is used as csip, sometimes as ip
15:14:48 <Xaroth> WEEEEEEEEEEEEEEEEEEEEEEEEEEEEEEEEEEEEEEEEEEEEEEEEEEEEEEEEEEEEEEEEEEEEEEEEEEEkend
15:15:28 <Xaroth> next week friday == karting with company
15:15:33 <Xaroth> week after that also a long weekend
15:15:41 <Xaroth> then vacation from xmas to after newyear
15:15:47 <Xaroth> == me collecting vacation hours :)
15:28:35 <DorpsGek> SVN: glx (r654) -Add: figured out a few variables
15:29:07 <DorpsGek> SVN: glx (r655) -Add: named another function
15:50:13 <Xaroth> there's a web API to initiate calls
15:50:30 <Xaroth> sooooo, i made a small script around it.. called 'bananaphone'
15:50:33 <Xaroth> to make somebody call himself
15:51:02 <Xaroth> in case of emergency, etc :)
15:51:43 <TrueBrain> or just to annoy them?
15:52:11 <Xaroth> if they piss me off with annoying conversations i'll just make them call themselves during that
15:52:19 <Xaroth> they'll rush to their phone, no more annoying conversations for me :)
15:53:07 <Xaroth> he spends 90% of his time facebooking/hyving/skyping to other people
15:53:38 <Xaroth> in between sentences in here.
15:54:17 <Xaroth> seriously tho, last month I spent 61 hours on tickets from the customer i'm assigned to (which is just 1 customer)
15:54:29 <Xaroth> he spent, 45 minutes on the remaining.. 8 ?
15:54:50 <Xaroth> and dares to exclaim that he's busy with tickets
15:55:35 <Xaroth> (of those 45 minutes he spent 30 minutes together with another colleague who did all the actual work for him..)
15:56:27 <Xaroth> so yes, I work, he doesn't
15:56:37 <Xaroth> so if he pisses me off, I put him back to work :)
15:58:45 <Xaroth> already tried that, but he apparently thinks he's doing a better job than he is
15:59:03 <Xaroth> so I'm waiting for him to fuck up (again)
16:04:05 <Xaroth> just 1 more hour to go until i got weekend
18:05:13 <glx> how useful, f__B491_0D4B_000D_85F6() first arg is unused
22:53:20 *** ChanServ sets mode: +v planetmaker
23:56:42 <TrueBrain> glx: that is to make things MUCH easier
continue to next day ⏵