| *** grantc has quit IRC | 00:50 | |
| *** KermitTheFragger has joined #ingres | 01:14 | |
| *** grantc has joined #ingres | 01:24 | |
| *** PaulM05 has joined #ingres | 01:43 | |
| PaulM05 | good morning all | 01:43 |
|---|---|---|
| grantc | morining PaulM05 | 01:46 |
| *** atrofast has joined #ingres | 02:27 | |
| atrofast | Everyone, how long does createdb take on your systems? | 02:28 |
| atrofast | I run Arch 64 bit and mine takes 40 secs | 02:28 |
| PaulM05 | what's the hardware? | 02:29 |
| *** Dejan has joined #ingres | 02:56 | |
| Dejan | \o/ | 02:56 |
| pboro | atrofast: | 02:58 |
| pboro | -bash-3.2$ time createdb testdb1 | 02:58 |
| pboro | real 0m0.670s, user 0m0.046s, sys 0m0.029s | 02:58 |
| pboro | the platform is RHEL 5.4 64-bit, 12 GB RAM, 15krpm disks | 02:58 |
| atrofast | PaulM05: Intel e8300 regular 7200rpm hdd | 02:58 |
| atrofast | Platform Linux ARch 64 bit | 02:58 |
| pboro | E5520 here, @ 2.27 GHz | 02:59 |
| atrofast | Which version of Ingres? | 02:59 |
| atrofast | I run SVN and Geospatial HEAD | 02:59 |
| Dejan | MESSAGE 'wassup fellas?' | 02:59 |
| pboro | atrofast, Ingres 9.2.0 with latest patch | 02:59 |
| atrofast | Everybody's createdb is faster than mine :( I'm getting an i5 750 today, gotta test it on that too | 03:00 |
| pboro | hehe :) | 03:00 |
| Dejan | atrofast, not really | 03:01 |
| Dejan | check this: | 03:01 |
| Dejan | real 0m8.402s | 03:01 |
| Dejan | user 0m0.040s | 03:01 |
| Dejan | sys 0m0.044s | 03:01 |
| atrofast | I have some suspicion that it's because I run 64 bit though because I also run a 32 bit fedora on the same machine and it's "only" 26 secs | 03:01 |
| Dejan | on my test server | 03:01 |
| atrofast | What's the hardware? | 03:01 |
| Dejan | quad core xeon | 03:01 |
| Dejan | and some scsii disks | 03:01 |
| Dejan | do not remember which ones | 03:01 |
| Dejan | it is IBM low-level server | 03:02 |
| Dejan | small one | 03:02 |
| atrofast | Ah so better than my hardware then :) | 03:02 |
| Dejan | Intel(R) Xeon(R) CPU X3320 @ 2.50GHz | 03:02 |
| atrofast | Does the Xeon do hyperthreadin? | 03:03 |
| Dejan | i think it does | 03:04 |
| grantc | atrofast, something is wrong slow with your disks/controllers | 03:08 |
| grantc | on a laptop i get it do create in 4.5 seconds | 03:08 |
| grantc | i assume iostat shows no activity? | 03:09 |
| Dejan | 4.5 sec ??? | 03:15 |
| Dejan | wow | 03:15 |
| grantc | dell latitude d620 running ubuntu 9.10 x86 and a 7.2k disk | 03:16 |
| atrofast | grantc: no, iostat shows about 150-200 TPS the whole time, and 2000-5000 blk_wrtn/s | 03:18 |
| atrofast | How high does your iostat jump on create? | 03:20 |
| atrofast | I used a 1 sec interval | 03:20 |
| Dejan | Device: tps Blk_read/s Blk_wrtn/s Blk_read Blk_wrtn | 03:22 |
| Dejan | sda 0.53 31.84 19.17 44098186 26552024 | 03:22 |
| grantc | line 55 in http://pastebin.com/f6bcbb29b gives you an idea | 03:25 |
| atrofast | Ah so my io is way down | 03:27 |
| atrofast | I did try with two different HDDs and about the same result... I'm trying on my laptop with teh same Arch 64 bit to see if it's any different | 03:27 |
| atrofast | Oh yeah, compiling Ingres on your laptop makes it hot :P | 03:30 |
| pboro | :D | 03:30 |
| *** Mud has quit IRC | 03:39 | |
| *** Mud has joined #ingres | 03:43 | |
| *** cthibert has joined #ingres | 04:11 | |
| *** zxiiro has quit IRC | 05:03 | |
| *** PaulM05 is now known as PaulM05_lunch | 05:08 | |
| atrofast | grantc: I did the same test on my laptop same OS but different hardware... Same result though createdb takes 45 secs | 05:09 |
| grantc | try a better OS then? | 05:09 |
| atrofast | But this is my OS of choice, it's Linux why would it be so much slower than other distros? | 05:11 |
| grantc | kernel (lack of) optimizations | 05:11 |
| atrofast | Ingres is the only thing that is slow :P | 05:11 |
| grantc | i've tried it on 9.04, 9.10 and suse 11.0 and its 1.9s, 4.5s and 0.9s respectively | 05:12 |
| grantc | first and last were on two different ESX servers (CA and Sluff) | 05:12 |
| grantc | what filesystem? | 05:13 |
| atrofast | Ext4, however Chuck and Andrew ran it on Ext4 with much better results | 05:14 |
| atrofast | Which version of INGres? I'm using SVN | 05:15 |
| grantc | if you try 9.2 for comparison? | 05:15 |
| grantc | i get 12.5s with 10.0 | 05:17 |
| atrofast | Heh I can't run it on Arch 64 bit, the binaries don't work properly for some reason | 05:17 |
| atrofast | On 64 or 32 bit system? | 05:17 |
| grantc | 32 | 05:18 |
| grantc | same as my earlier tests | 05:18 |
| grantc | it could be that optimzation is disabled in parts ? | 05:18 |
| atrofast | I'm building it pure 64 bit (without the 32 bit libs) | 05:18 |
| atrofast | Perhaps, but I haven't fiddled with that, so however Ingres builds it by default is what I use | 05:20 |
| grantc | it's worth raising a service desk ticket so that it gets some attention | 05:21 |
| grantc | a 400% slow down (in my case) ain't good | 05:21 |
| grantc | perhaps PaulM05 can reproduce the difference as well? | 05:22 |
| atrofast | Thanks for debugging with me grantc | 05:27 |
| *** rossand has joined #ingres | 05:47 | |
| *** ChanServ sets mode: +o rossand | 05:47 | |
| Dejan | grantc: am i allowed to install -com- package of Ingres, and continue using it after evaluation period expires ? | 05:48 |
| Dejan | does anyone know ? | 05:48 |
| grantc | IANAL but I would say no | 05:51 |
| Dejan | (i was thinking here for personal use) | 05:52 |
| grantc | i don't believe there are any technical measures to prevent you from going beyond the eval time but from a legal point I'd still say no. | 05:54 |
| Dejan | fair enuf | 05:54 |
| *** PaulM05_lunch is now known as PaulM05 | 05:58 | |
| PaulM05 | grantc - that 400% slowdown was that between different versions of Ingres, different OSs or 32v64? | 06:01 |
| *** zxiiro has joined #ingres | 06:02 | |
| *** ChanServ sets mode: +v zxiiro | 06:02 | |
| grantc | PaulM05, same machine / OS Ingres 9.2 vs 10.0 | 06:03 |
| PaulM05 | ok - let me check that | 06:04 |
| PaulM05 | 1.29s v 9.29s (9.2 v 10.0) | 06:16 |
| PaulM05 | not good | 06:16 |
| *** alwayslurking has joined #ingres | 06:17 | |
| *** alwayslurking has left #ingres | 06:18 | |
| atrofast | PaulM05: ANd I get a further 4 times slow down on my Arch 64 bit system with a Pure 64 bit build | 06:32 |
| PaulM05 | are you comparing that with a 32bit build on the same OS? | 06:33 |
| *** mull_ has joined #ingres | 06:37 | |
| Dejan | guys, what date alias type do you use ? | 06:40 |
| Dejan | ansi date, or ingres date ? | 06:41 |
| PaulM05 | ingresdate mostly | 06:44 |
| *** cthibert has left #ingres | 06:44 | |
| *** cthibert has joined #ingres | 06:54 | |
| grantc | Dejan, ingresdate but i use the ansi date types in schemas | 06:57 |
| atrofast | PaulM05: No I don't have a 32 bit install of this OS anywhere | 07:09 |
| PaulM05 | ok - so the "further 4 times slow down" relates to my figures? In which case it's not really meaningful unless we establish our respective hardware-OS combinations perform similarly | 07:10 |
| PaulM05 | but that's bye-the-bye as there's clearly an issue in 10.0 createdb | 07:10 |
| Dejan | i think ingres needs easer migration | 07:11 |
| Dejan | if i want to migrate ingres dbs from one host to another | 07:11 |
| Dejan | it takes ages to do everything | 07:11 |
| Dejan | one step towards easier migration would be some sort of administration tool that will export the whole current system into some sort of config file | 07:12 |
| Dejan | and use the same tool on a new machine to configure newly installed ingres there | 07:13 |
| grantc | if migrating from machine a to machine b you can do "unloaddb a::dbname" from machine b then run ./unload.ing | 07:13 |
| Dejan | that is what i do atm | 07:13 |
| Dejan | but i have lots of dbs | 07:13 |
| grantc | feel free to write the tool :P | 07:13 |
| PaulM05 | grantc - you beat me to it! ;) | 07:14 |
| grantc | :)) | 07:14 |
| Dejan | with mysql, i simply stop mysql, scp whole mysql tree to the new server | 07:14 |
| Dejan | and start mysql there | 07:14 |
| Dejan | as simple as that | 07:14 |
| Dejan | sure i need to edit my.cnf | 07:14 |
| Dejan | :D | 07:14 |
| grantc | you can do the same with Ingres with one caveat | 07:14 |
| Dejan | if host name changes, it does not work | 07:15 |
| Dejan | ingres gets crazy | 07:15 |
| grantc | set II_HOSTNAME to the old hostname | 07:15 |
| Dejan | aha, did not try that | 07:15 |
| Dejan | ty! | 07:15 |
| grantc | or s/oldhostname/`iipmhost`/ in config.dat and update II_GCNxx_LCL_VNODE | 07:16 |
| grantc | where xx is your installation code | 07:17 |
| Dejan | and yeah, ingres version is not the same... | 07:17 |
| Dejan | but yeah, thanks for info, i will try that just for fun | 07:17 |
| *** Mud has quit IRC | 07:38 | |
| *** Mud has joined #ingres | 07:43 | |
| grantc | if the ingres version is not the same then you might need to run upgrade db | 08:08 |
| *** Alex| has quit IRC | 08:31 | |
| *** monda07 has joined #ingres | 08:52 | |
| *** DarylM has quit IRC | 08:54 | |
| *** monda07 is now known as DarylM | 08:54 | |
| *** withdefault has quit IRC | 09:11 | |
| *** withdefault has joined #ingres | 09:12 | |
| *** ChanServ sets mode: +o withdefault | 09:12 | |
| *** bonsaikitten has quit IRC | 09:20 | |
| *** bonsaikitten has joined #ingres | 09:22 | |
| *** DrEeevil has joined #ingres | 09:22 | |
| *** bonsaikitten has quit IRC | 09:22 | |
| *** PaulM05 has quit IRC | 09:22 | |
| *** DrEeevil is now known as bonsaikitten | 09:26 | |
| *** bonsaikitten is now known as AmazingPudding | 09:26 | |
| *** AmazingPudding is now known as bonsaikitten | 09:26 | |
| Dejan | a newbie question | 09:43 |
| Dejan | when i export users from accessdb | 09:43 |
| Dejan | into an sql file | 09:43 |
| Dejan | what db should i use to create those users | 09:43 |
| Dejan | on a new ingres server | 09:43 |
| Dejan | ??? | 09:43 |
| Dejan | iidbdb ? | 09:43 |
| *** grantc has quit IRC | 09:46 | |
| *** KermitTheFragger has quit IRC | 09:49 | |
| *** Dejan has quit IRC | 10:06 | |
| *** zxiiro has quit IRC | 10:29 | |
| *** zxiiro has joined #ingres | 10:40 | |
| *** ChanServ sets mode: +v zxiiro | 10:40 | |
| *** Mud has quit IRC | 11:38 | |
| *** Mud has joined #ingres | 11:43 | |
| *** Mud is now known as Mud|afk | 12:26 | |
| *** Mud|afk is now known as Mud | 12:31 | |
| *** Mud has quit IRC | 12:36 | |
| *** withdefault has left #ingres | 12:49 | |
| *** zxiiro has quit IRC | 12:51 | |
| *** cthibert has left #ingres | 13:03 | |
| *** zxiiro has joined #ingres | 13:58 | |
| *** ChanServ sets mode: +v zxiiro | 13:58 | |
| *** bonro01 has left #ingres | 15:38 | |
| *** mull_ has quit IRC | 17:01 | |
| *** mull_ has joined #ingres | 17:53 | |
| *** rossand has quit IRC | 18:03 | |
| *** hopper75 has quit IRC | 18:03 | |
| *** atrofast has quit IRC | 18:04 | |
| *** rossand has joined #ingres | 18:06 | |
| *** ChanServ sets mode: +o rossand | 18:06 | |
| *** mull_ has quit IRC | 18:17 | |
| *** hopper75 has joined #ingres | 18:21 | |
| *** hopper751 has joined #ingres | 18:32 | |
| *** hopper75 has quit IRC | 18:33 | |
| *** rossand has quit IRC | 18:50 | |
| *** rossand has joined #ingres | 19:09 | |
| *** ChanServ sets mode: +o rossand | 19:09 | |
| *** DarylM has quit IRC | 19:10 | |
| *** cytrinox has quit IRC | 19:23 | |
| *** cytrinox has joined #ingres | 19:25 | |
| *** rossand has quit IRC | 19:35 | |
| *** Alex| has joined #ingres | 22:31 | |
| *** ChanServ sets mode: +o Alex| | 22:31 | |
| *** Alex| has quit IRC | 22:56 | |
| *** Alex| has joined #ingres | 23:18 | |
| *** ChanServ sets mode: +o Alex| | 23:18 | |
| *** Mud has joined #ingres | 23:41 | |
Generated by irclog2html.py 2.7 by Marius Gedminas - find it at mg.pov.lt!