*** rossand has quit IRC | 03:40 | |
*** Alex|off is now known as Alex| | 07:04 | |
*** grantc_offline is now known as grantc | 07:52 | |
*** withdefault1 has joined #ingres | 08:07 | |
pboro | ah... query of form "SELECT ... UNION SELECT ... ORDER BY col1" crashes dbms in 9.2 :) | 08:15 |
---|---|---|
pboro | the col1 only exists in the left side SELECT, not in both | 08:16 |
pboro | I'll open an issue | 08:16 |
MagnusG| | pboro, that sounds like a good idea | 08:17 |
pboro | my developer team is really good at finding new ingres bugs ;) | 08:17 |
*** Alex| is now known as Alex|off | 08:31 | |
*** Gerhard has joined #ingres | 08:34 | |
*** grantc is now known as grantc_offline | 08:34 | |
*** grantc_offline is now known as grantc | 08:34 | |
*** withdefault1 has quit IRC | 08:42 | |
*** Alex|off is now known as Alex| | 08:44 | |
grantc | Hi SlimeyPete | 09:07 |
*** rossand has joined #ingres | 10:05 | |
*** ChanServ sets mode: +o rossand | 10:05 | |
grantc | Ingres gets a mention in the guardian of all places - http://www.guardian.co.uk/technology/blog/2009/feb/24/opensource-internet | 10:52 |
grantc | SlimeyPete, I have done some testing with PHP 5.2.6 and 5.2.8 and various driver levels | 11:02 |
grantc | whilst going from 1 -> 2 does use more memory I don't see the bloat you report. | 11:03 |
grantc | However comparing 5.2.6 to 5.2.8 the former uses more memory | 11:03 |
grantc | for example with http://ingres.pastebin.com/m77fe75fd i get the following output with 5.2.6 (http://ingres.pastebin.com/m76771a6) and 5.2.8 ( | 11:08 |
grantc | http://ingres.pastebin.com/m31471b80 | 11:08 |
grantc | with the latest code in the 1.x branch - 5.2.6 - http://ingres.pastebin.com/m76771a6, 5.2.8 - http://ingres.pastebin.com/m64f246b4 | 11:11 |
*** withdefault1 has joined #ingres | 11:30 | |
*** withdefault1 has quit IRC | 11:33 | |
*** grantc has quit IRC | 11:37 | |
*** _withdefault has joined #ingres | 11:37 | |
*** _withdefault is now known as grantc | 11:38 | |
*** ChanServ sets mode: +o grantc | 11:38 | |
*** withdefault has joined #ingres | 11:55 | |
*** ChanServ sets mode: +o withdefault | 11:56 | |
withdefault | bhelp | 12:03 |
Alex| | :) | 12:03 |
withdefault | that's one for the log :) | 12:04 |
Alex| | hehe | 12:04 |
SlimeyPete | grantc: Hmm, okay, thanks | 12:39 |
* SlimeyPete scratches head and wonders if there's something a bit fundamentally weird with the server | 12:40 | |
grantc | it's certainly worth going with 5.2.8 | 12:40 |
grantc | note I also build in debug mode, for php and the driver which I would expect might make things worse | 12:41 |
*** troal01 has joined #ingres | 12:57 | |
troal01 | Hey if you have a database that is "inconsistent" is there any way you cna fix it? | 12:58 |
grantc | verify the data then mark it consistent | 12:58 |
Alex| | verifydb | 12:58 |
grantc | verifydb just flips a flag to say it's ok | 12:58 |
troal01 | Thanks all | 13:00 |
troal01 | It got inconsistent during a copy into statement.. Could that screw anything up badly? I m ean it's just reading tables, not writing | 13:04 |
grantc | if the database was not closed properly,yes | 13:05 |
troal01 | But it's just reading isn't it? :( | 13:07 |
grantc | if the process failed and the server still thinks it has a connection... | 13:07 |
grantc | why did the copy into fail? | 13:07 |
troal01 | I think my external HDD doesn't like really large file copies | 13:09 |
troal01 | Every time the binary file hit 77GBs it stopped writing | 13:09 |
Alex| | uh, not the best thing for a consistent database :) | 13:10 |
grantc | am not sure how you expect ingres to cope with your hardware being picky about what it writes :) | 13:10 |
troal01 | Yeah it's a mess, going to get a 1TB SATA HDD Friday | 13:10 |
troal01 | Hopefully it will work better | 13:11 |
grantc | i hope it's not one of those buggy segates | 13:11 |
troal01 | Heh | 13:11 |
troal01 | I haven't decided on brand | 13:11 |
troal01 | But I have 1 samsung and 2 seagates in my machine right now | 13:11 |
grantc | until the recent debarcle with freezing disks i would have gone with seagates | 13:12 |
SlimeyPete | Every major HD manufacturer has a catastrophe once every few years; it's all a bit of a gamble | 13:13 |
troal01 | So... what's a good affordable brand? | 13:13 |
grantc | perhaps but they botched the patch twice i believe | 13:13 |
grantc | "affordable" is a relative term | 13:13 |
Alex| | indeed.... while I'm still runnong one if IBM's DTLAs - the ones that were said to be a total failure | 13:13 |
SlimeyPete | Western Digital are pretty good. I usually go with them. | 13:14 |
pboro | I choose one which has the longest warranty, since they all break anyway | 13:14 |
SlimeyPete | but I have had WD drives fail before so they're certainly not invulnerable. | 13:14 |
Alex| | I guess it's in the nature of a spinning disk to fail at some point :) | 13:15 |
pboro | Alex|, agreed :) | 13:15 |
SlimeyPete | and in one case they managed to take a full 18 months to send me a new hard disk | 13:15 |
troal01 | That's a long wait | 13:15 |
SlimeyPete | it took so long that I'd forgotten all about it - all of a sudden a brand new hard disk arrived from Germany | 13:15 |
SlimeyPete | took me ages to figure out why they were sending me free hardware | 13:15 |
troal01 | I have a six year old 50GB Western Digital that's still running | 13:15 |
troal01 | Haha | 13:16 |
pboro | SlimeyPete :D | 13:16 |
withdefault | I had to replace a 400GB WD, luckily it was in warranty, but only just | 13:16 |
SlimeyPete | I reckon my broken disk had gotten put in a cupboard or at the back of a shelf, and then one day they moved offices or did a spring-clean and said "oh *crap*..." | 13:17 |
*** MagnusG| has quit IRC | 13:54 | |
troal01 | So... you guys don't think I should go for this one? http://oemexpress.com/Itemdesc.asp?ic=HDDSEA-3100034S&eq= | 14:09 |
grantc | http://www.channelregister.co.uk/2009/01/21/seagate_firmware_fix_breaks_barracudas/ | 14:21 |
troal01 | Ouch, that's bad | 14:42 |
troal01 | Firmware upgrades that brick your hardware = very bad | 14:43 |
*** DarylM has joined #ingres | 14:47 | |
troal01 | BUG! select first 10 * from main offset 10000000000;\g gives me an error | 15:06 |
troal01 | Says 10000000000 isn't an integer constant | 15:07 |
troal01 | I take it offset is an i4? | 15:07 |
Alex| | Doug will know | 15:08 |
pboro | it is possible to rollforwarddb only certain tables... is there something I should consider when doing this? | 15:10 |
DarylM | Don't do it unless you are desperate...... | 15:14 |
DarylM | but really... | 15:14 |
DarylM | referential integrity is your biggest concern | 15:14 |
pboro | ok, I'll do full rollforwarddb then | 15:14 |
DarylM | On production, yes. I most situations I would consider piecemeal disaster recovery too risky. | 15:16 |
grantc | troal01, log the bug on servicedesk, | 15:28 |
troal01 | Okay will do grantc | 15:28 |
*** Alex| is now known as Alex|off | 16:06 | |
*** Gerhard has quit IRC | 16:11 | |
grantc | ingres now on the bbc - http://news.bbc.co.uk/1/hi/technology/7910110.stm | 16:17 |
*** Alex|off is now known as Alex| | 16:18 | |
*** DerMeister has joined #ingres | 16:38 | |
*** grantc is now known as grantc_offline | 18:28 | |
*** troal01 has left #ingres | 18:50 | |
*** Alex| is now known as Alex|off | 19:07 | |
*** Gerhard has joined #ingres | 21:10 | |
*** Gerhard has quit IRC | 21:28 | |
*** withdefault is now known as notdefault | 21:57 | |
*** withdefault has joined #ingres | 21:59 | |
*** ChanServ sets mode: +o withdefault | 21:59 | |
*** withdefault has quit IRC | 22:00 | |
*** DerMeister has quit IRC | 22:16 | |
elPiola | grantc, I didn't realize that the UK has a "minister of digital engagement". Sounds like the Ministry of Information from the movie Brazil | 22:25 |
DarylM | You sure it isn't closer to the Ministry of Silly Walks? | 22:25 |
*** DarylM has quit IRC | 23:03 |
Generated by irclog2html.py 2.7 by Marius Gedminas - find it at mg.pov.lt!