IRC log for #koha, 2011-12-17

← Previous day | Today | Next day → | Search | Index

All times shown according to UTC.

Time Nick Message
00:09 melia left #koha
00:12 yhager joined #koha
02:00 yhager joined #koha
02:01 bg joined #koha
02:05 trea joined #koha
02:33 kmkale joined #koha
02:34 kmkale Namaskaar #koha
03:40 yhager joined #koha
04:04 bigbrovar joined #koha
04:12 kmkale joined #koha
04:41 bigbrovar joined #koha
04:43 edveal joined #koha
04:55 edveal joined #koha
05:30 edveal left #koha
05:36 cait joined #koha
05:57 cait hi #koha
08:15 mbalmer joined #koha
08:51 yhager joined #koha
09:34 mbalmer 50% of sources checked, 16 mysqlisms so far...
09:50 it would already help if proper standard sql was being used instead of mysql idioms.
09:54 cait I think that is a good goal
09:58 mbalmer step:  change SQL so that it can run on both mysql and pgsql
09:58 wahanui mbalmer: that doesn't look right
09:59 mbalmer wahanui, I am now up to 30 such spots...
09:59 wahanui ...but mbalmer is currently going each and every file to find MySQLisms, then I can think about how to make it Pg/MySQL "compliant"...
10:03 mbalmer yes.
10:04 cait hm
10:04 i hope someone has told you that wahanui is our bot
10:04 :)
10:05 hm if you are checking all sql currently... you could perhaps also check that placeholders are properly used everywhere?
10:07 mbalmer I will go trhough the SQL at least two times.  First to make it compatible, second to make it right where it's not
10:07 cait right?
10:07 mbalmer no, I did not know wahanui was a bot, but at least he is friendly ;)
10:07 cait most of the time
10:08 mbalmer as for right:  I found a few places where sth could be achieved with a more elegant sql statement
10:08 but it's a large codebase, it takes time to go through it.
10:16 cait yeah, it has grown a lot over the years
10:19 mbalmer quite some of the mysqlisms can be rewritten using standard SQL, so that's low hanging fruit
10:21 I need to setup a wiki page with info on that.
10:24 cait I think that would be good
10:25 also help people to not make the same mistakes again
10:25 ok, have to leave, will be back later today
10:41 kmkale joined #koha
11:32 cait joined #koha
11:38 cait hey magnuse
11:38 waking up now? :)
12:44 jcamins_away Heh. magnuse understands sleeping in on weekends. :)
12:44 cait apparently :)
12:44 hi jared :)
12:45 jcamins_away Good morning.
12:49 jcamins Myshkin says hi.
12:52 * cait waves back at Myshkin
17:00 julian_m joined #koha
17:33 adminacct joined #koha
17:33 indradg server sizing for Koha?
17:34 koha hardware requirement?
17:34 jcamins indradg: not sure wahanui has anything about that.
17:34 memory for koha?
17:34 indradg jcamins, ah well :)
17:35 francharb joined #koha
17:35 indradg jcamins, kinda... i had proposed a config for a koha server... and now
17:35 its come back to haunt me
17:36 francharb hi
17:36 wahanui bonjour, francharb
17:36 jcamins indradg: are you trying to figure out how large a server you need, or were you just wondering what wahanui had to say?
17:36 Uh-oh. What did you suggest?
18:24 druthb joined #koha
18:26 druthb o/
18:27 jcamins \o
18:27 cait hi druthb :)
18:28 druthb :D
18:58 stuartyeates joined #koha
19:04 indradg joined #koha
19:06 jcamins Welcome back, indradg.
19:07 indradg jcamins, sorry was on phone... these were libs with typically 25k in collections each with about 500 - 1200 members, so i had placed a Intel Core 2 Duo E7600, 3.06 GHz with 2gigs of RAM as the server
19:07 jcamins, thanks :)
19:07 jcamins indradg: I would've thought that would be plenty.
19:08 * druthb concurs;  that actually sounds like overkill, to me.
19:08 jcamins I guess I would've thought wrong?
19:08 druthb: well, with that many patrons for such a small library, I can see some pretty heavy circ usage.
19:09 druthb Even if all of the members came in twice a day, you're still not looking at 10K transactions a day, jcamins.
19:09 indradg druthb, jcamins i had to factor another thing.. these are govt libs... and they wont be able to upgrade in next 4 years... so yeah overbuilt the boxes a bit
19:09 jcamins druthb: true.
19:10 indradg the server would additionally function as their web proxy with content filter and dhcp server for their 3 node LANs
19:10 druthb yeah, sounds like, and that's good.  Should provide plenty of growth room, too, unless you put a teeny tiny hard drive on em
19:10 indradg plus offer web-dav shared folders
19:11 the servers went with 320 GB SATA HDD
19:11 druthb for the Koha part, that should be puh-lenty.
19:11 indradg the boxes are essentially PC systems
19:11 druthb so what ended up being the choke-point?
19:12 jcamins druthb: I saw 5000-12000 members.
19:12 Just realized I added a zero.
19:12 druthb ;-)
19:13 indradg we did the rollout at 170 libraries... and now there is a new govt and want to know if there are any "koha documentation" as far as h/w sizing estimates are concerned
19:13 jcamins indradg: 170 libraries? Wow!
19:13 druthb ah-hah.  so, not a technological sticking point, but a *political* one.
19:13 indradg i believe zebra can consume upto 500 MB ? jcamins druthb ?
19:13 druthb, yep... precisely
19:13 jcamins In terms of RAM?
19:14 I've never seen Zebra hit the default maxmem limit.
19:14 And I think that's a lot lower.
19:14 druthb I've been meaning to do some math to correlate memory/disk vs koha size (transaction load, bib and borrower counts), but have not gotten that done.
19:14 indradg jcamins, yeah... i vaguely remember seeing something that zebra can take up to 500 MB or so
19:14 * druthb has a great long list of libraries to help provide data.
19:14 jcamins indradg: you're off by a zero.
19:15 rangi never seen it get about about 30meg
19:15 jcamins *50* MB.
19:15 indradg rangi, jcamins thankee!
19:15 jcamins Apparently my inability to read is contagious. ;)
19:15 indradg: in terms of disk space, though, Zebra can take a fairly large amount.
19:16 * druthb squees and waves to rangi
19:16 rangi heya druthb :)
19:16 jcamins rangi: how large is the largest Zebra DB you've worked with?
19:16 rangi 10 gig
19:16 that was 3 million bibs
19:16 jcamins rangi: is that the space taken up by indexes, or the space taken up by the raw MARC data?
19:17 rangi indexes
19:17 raw marc is well less than that
19:17 indradg i find this on the m/l archives... but dunno if he is talking about the system in general or zebra in particular - http://lists.koha-community.or[…]March/032032.html
19:18 jcamins Huh. I think the disk space requirements for Zebra indexes must not scale linearly.
19:18 indradg: he's talking about total RAM.
19:18 indradg hmm..k
19:18 jcamins I'd agree with that.
19:18 rangi yeah i dont see that
19:18 jcamins rangi: you don't think so?
19:19 rangi: I wouldn't want to run a production system with <512MB of RAM at this point.
19:19 rangi i thought you were talking zebra
19:19 but no
19:19 druthb that's way conservative.    I've got a DB with 300K+ bibs that doesn't take up more than 7 or 8 GB for the mysql data.
19:19 rangi 512 ram is ridiculous
19:19 zebra would be fine with that much
19:19 apache and mysql no
19:20 jcamins Ah.
19:20 Yeah, agreed.
19:20 With both points.
19:21 rangi apache is the ram hog
19:21 jcamins indradg: one point about the disk use for Zebra indexes: rangi is talking about the disk used by a fully-indexed database. If you're going to want the option of reindexing without totally taking down the system, you need to double the disk space requirement.
19:21 rangi if you dont set up your max connections and threads
19:22 yeah but double 10 is still small
19:22 indradg i get tat
19:22 that*
19:22 jcamins rangi: true.
19:22 rangi: unless you're on Rackspace.
19:22 rangi yeah
19:22 cloud sux
19:23 jcamins EC2 is nice, but Rackspace...
19:23 * jcamins shudders
19:23 rangi nice until it goes down for 12 hours
19:23 its only nice if you have dr
19:23 cait dr?
19:24 rangi disaster recovery
19:24 jcamins Disaster recovery.
19:24 rangi ask scls about that
19:24 indradg so do we have any docs on the wiki about tentative h/w sizing guidelines? i tried to search... didn't find any
19:25 rangi nope
19:25 jcamins indradg: I don't think so.
19:25 * jcamins encourages you to write one. :)
19:25 rangi because to do so, is usually not that useful
19:25 its usually a how long is a piece of string
19:25 question
19:26 * druthb notes that RFPs are usually chock-full of such useless questions.
19:26 jcamins rangi: unless some bureaucrat wants to check off a box saying that the document exists, in which case it doesn't matter, it just needs to be there for box checking.
19:27 rangi problem is, someone will take it seriously :)
19:27 * jcamins suggests a document staying "Hardware should be no less than is required, but no more than is necessary."
19:27 rangi there ya go
19:27 jcamins s/staying/saying/
19:27 indradg well... for myself I do not need that doc... but these govt types are picking my brains...
19:28 rangi yup, write one, and put it on the wiki and refer to that :-)
19:28 * druthb suggests a wiki template tag that puts in "This document exists solely so that bureaucrats can check off boxes.  don't take this seriously in the real world, ever."
19:28 rangi :-)
19:28 indradg druthb, ++
19:28 rangi maybe slightly diplomatically worded
19:29 'This exists purely to satisfy RFP type requirements, please do your own research'
19:29 type thing
19:29 druthb A tag at the bottom, maybe, that says, "if you're taking this seriously, you must not actually do this for a living?"
19:29 rangi heh
19:34 * jcamins added 300 words worth of footnotes.
19:36 jcamins Yikes.
19:37 cait sounds very academic :)
19:42 jcamins cait: I guess I need to rewrite the rest of it so the rest is as academic-sounding. :)
20:17 bag @wunder 93109
20:17 huginn` bag: The current temperature in K6LCM - Westside / Mesa, Santa Barbara, California is 16.0°C (12:17 PM PST on December 17, 2011). Conditions: Overcast. Humidity: 54%. Dew Point: 7.0°C. Pressure: 30.03 in 1016.8 hPa (Falling).
21:52 samn joined #koha
22:03 cait left #koha
22:46 Soupermanito joined #koha

← Previous day | Today | Next day → | Search | Index

koha1