*** Alex|off is now known as Alex| | 00:18 | |
*** grantc has joined #ingres | 00:45 | |
*** ChanServ sets mode: +o grantc | 00:45 | |
*** Deyan has joined #ingres | 02:10 | |
Deyan | hello guys | 02:10 |
---|---|---|
pboro | hi Deyan | 02:26 |
*** thiagomz has quit IRC | 03:06 | |
*** rossand has joined #ingres | 03:07 | |
*** ChanServ sets mode: +o rossand | 03:07 | |
Deyan | how's going, guys? | 03:11 |
pboro | fine :) I installed patch 13457 to fix a bug with JDBC and LOBs, but found a new bug... | 03:12 |
Deyan | do i have to do some special configuration in order to build Ingres on x86_64 Linux? | 03:16 |
Deyan | when i do what README says i get bunch of errors at some point | 03:16 |
pboro | what are the errors? paste them in pastebin and paste url here | 03:17 |
Deyan | the first problem i see is: "don't know how to make /usr/lib/libxerces-c.so.27.0" | 03:19 |
Deyan | which is stupid | 03:19 |
Deyan | because i have libxerces-c | 03:19 |
pboro | ah, that has been discussed before | 03:19 |
Deyan | in /usr/lib64 | 03:19 |
pboro | you need 32-bit version of libraries too | 03:19 |
Deyan | wtf? | 03:19 |
Deyan | o.O | 03:20 |
pboro | since Ingres build is a hybrid build, it builds both 32- and 64-bit versions | 03:20 |
pboro | https://help.ubuntu.com/community/Servers/Ingres | 03:22 |
pboro | there's tips for compiling ingres on Ubuntu | 03:22 |
pboro | it's written by atrofast, who's also on the channel | 03:23 |
pboro | the tips mention Xerces 2.5 but based on your error, it expects 2.7 nowadays | 03:25 |
Deyan | i have 2.8 | 03:25 |
Deyan | on the system | 03:25 |
Deyan | and i do not have 32bit libs at all | 03:25 |
atrofast | Deyan: You will need 32 bit libs to build Ingres on a 64 bit machine | 03:53 |
atrofast | They vary somewhat between systems and packages but you will need a GCC that has both 64 and 32 bit libs, and xerces-c needs to be there, both 32 bit and 64 bit | 03:53 |
*** dyki has joined #ingres | 03:55 | |
dyki | grantc: what is the maximum inserts that php can handle in single transaction? | 03:56 |
dyki | number of^^ | 03:56 |
pboro | I don't think PHP puts any limit on it... but your transaction log size will :) | 03:56 |
grantc | dyki, what pboro says | 03:57 |
pboro | dyki, all data goes to transaction log until you commit, so you have to make sure you don't fill the log or make the log big enough | 03:58 |
grantc | an alternative is to use savepoints, but the driver does not support them, yet | 03:59 |
dyki | thanks guys | 04:00 |
dyki | i couldnt figure out why my script failed to commit when it had more than 300 rows to insert. | 04:01 |
*** rossand has quit IRC | 04:04 | |
*** rossand has joined #ingres | 04:14 | |
*** ChanServ sets mode: +o rossand | 04:14 | |
*** rossand has quit IRC | 04:45 | |
pboro | grantc, http://svn.ingres.com/ | 04:48 |
pboro | you might want to replace that with a redirect to www.ingres.com or similar ;) | 04:48 |
grantc | red is quite fashionable these days :) | 04:48 |
Alex| | oh yes indeed... :) | 04:49 |
grantc | another "reload" required | 04:49 |
pboro | Ah, my ingres build environment works now. I finally decided to make few enhancements to certain error messages... | 04:51 |
grantc | sorted http://svn.ingres.com/ goes to http://svn.ingres.com/svn | 04:53 |
Deyan | grantc: does ingres extension automatically quote strings? | 05:04 |
grantc | what do you mean by quote? | 05:05 |
Deyan | i have something like: "insert into ... values(?, ?, '?'...)" | 05:05 |
Deyan | if i have single quotes there ingres_query does not understand it | 05:05 |
Deyan | if i remove it, it does not generate proper statement | 05:05 |
grantc | if you are parsing params using ? then quoting strings is not needed | 05:06 |
Deyan | i am doing something wrong, but do not know what | 05:06 |
grantc | how are you passing parameters? using an array? | 05:06 |
Deyan | yes | 05:06 |
grantc | then you should not need to quote the parameter values | 05:07 |
grantc | the driver speaks to the server to work out what is being passed (ingres 9.1.x or newer) else it infers it from the PHP types | 05:07 |
grantc | or you can type cast in php | 05:07 |
Deyan | ty, something is wrong, but i do not see what | 05:08 |
Deyan | if i remove quotes i get bunch of different error messages :D | 05:09 |
Deyan | i'll make a proper strings out of it | 05:09 |
Deyan | maybe that will fix the problem | 05:10 |
Deyan | i am basically passing result of ingres_query | 05:10 |
Deyan | (elements of that array) | 05:10 |
Deyan | to the prepared statement query | 05:10 |
Deyan | PHP Warning: ingres_query(): Error binding parameters in /home/ingres/dejan/athena/migrator/inc/dir_dat_companies.php on line 251 | 05:11 |
Deyan | yeah, if i leave plain ? something goes wrong | 05:11 |
Deyan | let me make strings out of those... | 05:12 |
Deyan | and see if it will work | 05:12 |
grantc | does ingres_error() return anything? | 05:13 |
Deyan | it prints empty string | 05:20 |
grantc | looking at the code it's possible that the function exits without generating a message, apart from the one you see | 05:21 |
grantc | the function that does the bind being php_ii_bind_params | 05:22 |
*** mull has joined #ingres | 05:24 | |
*** toumi01 has joined #ingres | 05:59 | |
*** Alex| is now known as Alex|off | 06:41 | |
*** rossand has joined #ingres | 07:02 | |
*** ChanServ sets mode: +o rossand | 07:02 | |
atrofast | grantc: This FAQ talks a little about Trac + remote SVN repos: http://trac.edgewall.org/wiki/TracFaq#can-i-use-trac-with-a-remote-subversion-repository | 07:36 |
grantc | atrofast, thanks i read something a long those lines in someone's blog | 07:37 |
grantc | i think for now it's a no-no | 07:37 |
atrofast | Potentially you could mount the svn repos as a remote filesystem using sshfs or something but that doesn't seem like a very good idea | 07:38 |
atrofast | And it still wouldn't take care of the SVN hooks issue | 07:38 |
* grantc thinks atrofast should stop digging :) | 07:39 | |
atrofast | Hehe, okay... I think it *could* be done, not sure it *should* | 07:39 |
grantc | indeed :) | 07:40 |
grantc | perhaps we could move the ingres instance off this box on to another machine | 07:40 |
* Deyan loves redmine more | 07:40 | |
Deyan | :) | 07:40 |
Deyan | http://www.redmine.org | 07:40 |
grantc | that won't impact lxr or trac | 07:41 |
grantc | rails .... -1 | 07:41 |
atrofast | That would be an easier move, but what would we accomplish with that? | 07:42 |
grantc | moving ingres off would reduce the load and memory footprint | 07:46 |
Deyan | lighttpd ftw | 07:49 |
grantc | Deyan, lighttpd does not support ldap groups ... | 07:50 |
atrofast | Without crawlers on Trac and LXR I don't think Ingres is affecting the load and memory footprint that much o nthe machine. Although I don't have any hard data to back this up | 07:52 |
*** adrianwarwickuni has joined #ingres | 07:53 | |
grantc | hi adrianwarwickuni | 07:53 |
adrianwarwickuni | hello | 07:53 |
grantc | if you still have a DMF Q you might want to ask toumi01 | 07:54 |
adrianwarwickuni | Thanks. I've received a suggestion from Doug Inkster that I'm gonna try first. | 07:54 |
grantc | fair enough | 07:55 |
grantc | atrofast, my problem with how code.ingres.com is setup is that we don't have a good handle for what is going on | 07:56 |
grantc | performance/memory/networking | 07:57 |
atrofast | That's true, well moving Ingres to another server wouldn't be a big change, just update the connection info for LXR and Trac | 08:06 |
grantc | what do you think about moving / to /svn | 08:09 |
Deyan | brb - have to reboot | 08:09 |
*** Deyan has quit IRC | 08:09 | |
grantc | k | 08:09 |
atrofast | Think it would be fairly painless to use a rewrite to make all current requests go to /svn or would we have to inform everyone that they would have to do an svn switch? | 08:11 |
grantc | plan A is the re-write else it's plan B, switching | 08:12 |
*** withdefault has joined #ingres | 08:14 | |
*** ChanServ sets mode: +o withdefault | 08:14 | |
*** Alex|off is now known as Alex| | 08:51 | |
*** Deyan has joined #ingres | 08:54 | |
Deyan | meep | 08:55 |
Deyan | guys, is transaction log a file? | 09:01 |
Deyan | or a chunk of memory reserved for transactions? | 09:01 |
Deyan | CBF says it is a file | 09:01 |
Deyan | and is 32M big | 09:01 |
Deyan | by default | 09:01 |
pboro | it is file by default | 09:01 |
pboro | 32M is pretty small | 09:01 |
Deyan | if it is a file i can reserve 1G for it | 09:02 |
Deyan | or more | 09:02 |
pboro | sure | 09:02 |
pboro | I have 512 MB on my servers | 09:02 |
Deyan | i wish someone writes some good article how to properly configure Ingres after the instalation | 09:02 |
Deyan | because defaults are not good | 09:02 |
Deyan | it would help Ingres newbies like me | 09:03 |
Deyan | Ingres documentation is not helpful | 09:03 |
Deyan | it is too big | 09:03 |
atrofast | Deyan, you can use CBF to increase the size of your transaction log | 09:03 |
Deyan | atrofast: i am doing that atm | 09:03 |
atrofast | Okay, excellent... Yeah a 1 gig file is a lot better than 32mb .. I often make it a couple of gigs in case... Although the bigger it is the longer it can take for transactions to unwind or for the recovery server to start if you have any kind of crash or interruption | 09:04 |
Deyan | atrofast: any other hints to make Ingres perform better (after the installation)? :) | 09:05 |
Deyan | this server has 8GB RAM | 09:05 |
Deyan | I usually setup query cache(s) in MySQL | 09:05 |
Deyan | to fit the amount of RAM | 09:05 |
Deyan | and similar stuff | 09:06 |
pboro | MySQL's query cache sucks btw | 09:06 |
pboro | key_buffer_size is the most useful | 09:06 |
pboro | in Ingres I would make DMF caches at least "large" or even bigger with that amount of RAM | 09:07 |
Deyan | thanks | 09:07 |
Deyan | i am going to do that | 09:07 |
pboro | (cache_guideline or what ever the name of the derived settings was) | 09:07 |
Deyan | i am also going to read what DMF is :D | 09:07 |
Deyan | lol | 09:07 |
Deyan | there are several dmf_* parameters | 09:08 |
Deyan | dmf_db_cache_size = 40 entries | 09:09 |
Deyan | how big is "entry" ? | 09:09 |
pboro | DMF cache caches pages from database tables | 09:09 |
pboro | hmm I would recommend tuning cache_guideline instead of touching separate values | 09:09 |
Deyan | khm | 09:10 |
Deyan | where is that? | 09:10 |
pboro | lemme check the name of the setting | 09:10 |
Deyan | DBMS server parameters? | 09:10 |
Deyan | (9.3 Ingres here) | 09:10 |
pboro | there's separate DMF caches for each page sizes (2 kB, 4 kB, 8 kB, 16 kB, ...) | 09:10 |
pboro | cbf -> DBMS Server / Configure (1) -> Cache (4) | 09:11 |
Deyan | ok | 09:11 |
Deyan | going there | 09:11 |
Deyan | i just have ON/OFFs there | 09:12 |
pboro | in cache view there's listing like "DMF Cache 2k", "DMF Cache 4k", ... | 09:12 |
pboro | yeah, then choose 2k and Edit(2) | 09:12 |
pboro | ah sorry | 09:12 |
pboro | not edit, but Configure (1) | 09:12 |
pboro | edit switches on/off :) | 09:12 |
pboro | and in configure there's setting "cache_guideline" | 09:12 |
Deyan | it is "medium" here | 09:12 |
Deyan | everywhere | 09:13 |
pboro | yup... possible values are "small", "medium", "large" and "huge" | 09:13 |
Deyan | hell | 09:13 |
Deyan | searching documentation what it means... | 09:13 |
pboro | I have it configured "large" for all page sizes that are ON | 09:13 |
pboro | PF2 is help for the field, at least in vt100f | 09:14 |
pboro | or F2, depends on your terminal settings | 09:14 |
pboro | each of small/medium/large/huge takes different amount of memory, it is also possible to make cache too large, i.e. it is never used completely and the memory is wasted in that case | 09:15 |
Deyan | ingres uses very little memory atm | 09:15 |
Deyan | which is normal | 09:15 |
Deyan | it is a test server and has just few concurrent users | 09:16 |
pboro | yup | 09:16 |
pboro | hmm I wonder if there's a good web page about DMF caches that you could just read and understand... | 09:16 |
Deyan | if you find it, i will be eternaly grateful | 09:16 |
Deyan | :D | 09:16 |
pboro | well, at least you can google about DMF, I gotta run now, my gf wants to go eat :D | 09:17 |
pboro | tracepoint dm420 can be used to check dmf cache status, google dm420 and dmf or similar | 09:17 |
pboro | now -> | 09:17 |
Deyan | where are Ingres headquarters? | 09:34 |
Deyan | is* | 09:35 |
Deyan | USA or UK? | 09:35 |
grantc | California | 09:35 |
Deyan | kewl | 09:35 |
grantc | but the EU HQ is in the UK | 09:37 |
*** Deyan has quit IRC | 10:03 | |
*** withdefault has left #ingres | 10:14 | |
*** Gerhard has joined #ingres | 10:35 | |
*** grantc has quit IRC | 11:14 | |
*** Gerhard has quit IRC | 11:41 | |
*** atrofast has left #ingres | 11:46 | |
*** Alex| is now known as Alex|off | 11:56 | |
*** rossand has quit IRC | 12:17 | |
*** DerMeister has joined #ingres | 12:19 | |
*** Dejan has joined #ingres | 13:10 | |
Dejan | hi everybody | 13:12 |
*** rossand has joined #ingres | 13:13 | |
*** ChanServ sets mode: +o rossand | 13:13 | |
*** Dejan has quit IRC | 13:44 | |
*** Dejan has joined #ingres | 13:52 | |
*** Dejan has quit IRC | 13:54 | |
*** DerMeister has quit IRC | 14:21 | |
*** rossand has quit IRC | 14:31 | |
*** rossand has joined #ingres | 14:32 | |
*** ChanServ sets mode: +o rossand | 14:32 | |
*** mull has quit IRC | 15:42 | |
*** adrianwarwickuni has quit IRC | 16:16 | |
*** rossand has quit IRC | 16:52 | |
*** zxiiro has quit IRC | 21:57 | |
*** zxiiro has joined #ingres | 21:57 | |
*** ChanServ sets mode: +v zxiiro | 21:57 | |
*** Alex|off is now known as Alex| | 22:38 | |
*** grantc has joined #ingres | 23:03 | |
*** ChanServ sets mode: +o grantc | 23:03 | |
*** Alex| is now known as Alex|off | 23:07 | |
*** Alex|off is now known as Alex| | 23:23 |
Generated by irclog2html.py 2.7 by Marius Gedminas - find it at mg.pov.lt!