*** rossand has joined #ingres | 02:44 | |
*** ChanServ sets mode: +o rossand | 02:44 | |
Dejan | ultrav1olet: sql <database name> < <sql file> | 03:13 |
---|---|---|
Dejan | example: sql mydb < commands.sql | 03:14 |
Dejan | or: cat commands.sql | sql mydb | 03:14 |
Dejan | i think there is also a \script command | 03:15 |
ultrav1olet | Dejan: got that | 03:15 |
ultrav1olet | right now my ingres server segfaults after start up | 03:16 |
ultrav1olet | I wiped transaction log since it got full | 03:16 |
ultrav1olet | how can I recreate all ingres work files? | 03:16 |
ultrav1olet | without reinstalling it? | 03:16 |
Dejan | increase size of the LOG file | 03:16 |
Dejan | by default it is very small | 03:17 |
Dejan | like 256Mb | 03:17 |
Dejan | or something like that | 03:17 |
ultrav1olet | right now it's 2G | 03:17 |
Dejan | aha, that is good | 03:17 |
ultrav1olet | right now my ingres server segfaults after start up, how can I recreate all ingres working files? | 03:17 |
Dejan | what does it say? | 03:17 |
ultrav1olet | 200KB of errors | 03:18 |
Dejan | just see the last error | 03:19 |
Dejan | that would where crash-related errors will appear | 03:19 |
Dejan | ie. the end of file | 03:19 |
ultrav1olet | that is iicrash.log | 03:19 |
Dejan | no, error.log | 03:19 |
Dejan | iicrash is impossible to understand by humans :D | 03:20 |
ultrav1olet | Fri Oct 23 01:01:26 2009 E_DMA460_LGREAD_EOF_REACHED The end of th | 03:20 |
ultrav1olet | e log file has been reached. Log record address <1256284382,22,0> is past the end (or beginning) of the log file. The log file currently contains log record | 03:20 |
ultrav1olet | addresses <1256284382,26,120> through <1256284382,32,740>. | 03:20 |
Dejan | so log file is full again | 03:22 |
ultrav1olet | http://pastebin.ca/1639706 | 03:22 |
ultrav1olet | Dejan: transaction log or log file? | 03:22 |
ultrav1olet | I have recreated transaction log and did nothing after that. It cannot be full | 03:23 |
Dejan | that is what buffles me | 03:24 |
Dejan | are you sure you restarted ingres? | 03:24 |
ultrav1olet | yes, of course | 03:25 |
ultrav1olet | checked it wasn't running with ingstatus | 03:25 |
ultrav1olet | maybe I shouldn't work with ingres-10.0.0-107-NPTL-gpl-pc-linux-i386-LONG.tgz | 03:25 |
Dejan | that is not good | 03:25 |
Dejan | i check with ps -aux | 03:25 |
Dejan | :) | 03:25 |
ultrav1olet | but ingres-10.0.0-109-NPTL-gpl-pc-linux-i386.tgz is broken | 03:25 |
Dejan | i do not trust ingstatus | 03:25 |
ultrav1olet | ps axu showed zero ingres processes | 03:25 |
Dejan | i use build 109 atm | 03:25 |
Dejan | ok | 03:26 |
ultrav1olet | it has a broken RPM inside | 03:26 |
Dejan | then i have no idea how to help you | 03:26 |
grantc | ultrav1olet: that release of ingres is very much an alpha release - it has partial support for longer names | 03:26 |
Dejan | i am an Ingres newbie myself | 03:26 |
ultrav1olet | grantc: which one should I use, then? | 03:26 |
ultrav1olet | ingres-9.3.0-151-NPTL-gpl-pc-linux-i386.tgz ? | 03:26 |
Vroomfondle | 9.3 is the latest full release. 9.2 is the more "mature" option. | 03:27 |
grantc | what Vroomfondle says | 03:27 |
ultrav1olet | ok, let's purge 10 | 03:27 |
grantc | 9.3 just got released - v10 is still being developed | 03:28 |
grantc | eg last id support went in yesterday/today ... | 03:28 |
Dejan | omg, really? | 03:30 |
Dejan | that is good to "hear" | 03:30 |
Dejan | better say read... | 03:30 |
Dejan | :) | 03:30 |
ultrav1olet | I'm going to have a 20GB database, what is the proper log size for importing such a db? | 03:30 |
grantc | ultrav1olet: it depends ... | 03:33 |
ultrav1olet | transaction log | 03:33 |
ultrav1olet | well, with the default transaction log ingres failed importing 100MB database. | 03:33 |
ultrav1olet | that's ridiculous | 03:34 |
Dejan | strange | 03:34 |
grantc | if you are importing each table from scratch you can make use of bulk load which will avoid some of these problems | 03:34 |
Dejan | i am importing 3GB databases like 40 times per day | 03:34 |
Dejan | and i have 2G transaction log | 03:34 |
Dejan | however, when i try larger databases, ingres crashes | 03:35 |
Dejan | so what i did is simply to split copy.in into chunks | 03:35 |
Dejan | and i execute them with 10sec pause in between | 03:36 |
Dejan | as i am a newbie i have no idea why, but ... it is working ;) | 03:36 |
grantc | http://docs.ingres.com/Ingres/9.3/Database%20Administrator%20Guide/bulkcopy.htm#o1959 | 03:37 |
Dejan | i have a decade of work with MySQL behind me, but when it comest to Ingres, i am still a noob | 03:37 |
ultrav1olet | same here | 03:37 |
grantc | if you don't meet the conditions for a bulk copy we do an incremental copy and the trx log file gets used a lot more | 03:37 |
Dejan | grantc: good to know that, thx | 03:38 |
grantc | you can load the database using "set nologging" but that is not advisable | 03:38 |
ultrav1olet | I'd be glad to disable it altogether | 03:38 |
ultrav1olet | great! | 03:38 |
ultrav1olet | is this just an SQL statement before running sql db < data.sql ? | 03:39 |
grantc | "set nologging" will disable it but it will make your database inconsistent if there is any problem | 03:39 |
Dejan | ultrav1olet: put it as the first line in data.sql | 03:39 |
ultrav1olet | Dejan: that's exactly what I thought | 03:39 |
grantc | you would put "set nologging" at the top of your script and "set logging" at the end to re-enable | 03:39 |
grantc | see http://docs.ingres.com/Ingres/9.3/Database%20Administrator%20Guide/largedataloadswsetnologgingstatement.htm for the "set nologgin" statement | 03:41 |
Dejan | yeah | 03:42 |
Dejan | and make a checkpoint after | 03:42 |
Dejan | :D | 03:42 |
ultrav1olet | damn, The transaction log file is full. The transaction will be aborted. | 03:49 |
ultrav1olet | that's with set nologging | 04:05 |
*** ultrav1olet has left #ingres | 04:10 | |
Dejan | select * from ii_log where nickname = 'Dejan';\g | 04:44 |
Dejan | grantc: i just got a good idea | 04:44 |
Dejan | for a bot | 04:44 |
Dejan | :D | 04:44 |
grantc | go on.. | 04:45 |
Dejan | what you think about bot which actually can execute SQL statements | 04:45 |
Dejan | some sort of IRC frontend for TM | 04:45 |
* grantc thinks that's dangerous | 04:45 | |
Dejan | each nickname here on this channel would get a schema | 04:45 |
grantc | why would you need it? | 04:46 |
Dejan | with enough permissions to do common SQL stuff | 04:46 |
Dejan | to learn SQL | 04:46 |
Dejan | and Ingews | 04:46 |
Dejan | upsy, Ingres | 04:46 |
Dejan | it is not about learning SQL | 04:46 |
Dejan | but about discussions | 04:47 |
Dejan | it would be similar to what #python, #minid, #lua guys have | 04:47 |
Dejan | on their channels | 04:47 |
Dejan | #D too | 04:47 |
grantc | i'll have a think about it... | 04:51 |
*** mull has joined #ingres | 06:30 | |
*** grantc has quit IRC | 06:48 | |
Dejan | are there going to be MD5() or SHA1() functions in Ingres? | 08:24 |
*** Dejan has quit IRC | 08:31 | |
*** DerMeister has joined #ingres | 08:40 | |
*** Dejan has joined #ingres | 09:33 | |
Dejan | hello | 09:33 |
Dejan | anyone here? | 09:36 |
pboro | sure | 09:36 |
Dejan | pboro: do you perhaps know if it is planned to have MD5 or SHA1 functions in Ingres? | 09:38 |
pboro | haven't heard about them, but you can always implement them yourself | 09:38 |
Dejan | can i? - i think i must read some documentation about it | 09:48 |
pboro | http://community.ingres.com/wiki/OME:_User_Defined_Functions | 09:49 |
Dejan | YAY | 09:50 |
Dejan | IT HAS MD5 sum as an example! | 09:50 |
Dejan | wooohoooo | 09:50 |
pboro | yup | 09:50 |
Dejan | page bookmarked | 09:53 |
Dejan | ty | 09:53 |
pboro | np | 09:53 |
Dejan | i wonder if Ingres developers ever thought of making this even easier | 09:53 |
Dejan | to actually have some sort of "repository" with all "commands" | 09:54 |
Dejan | that are in fact simple linux apps | 09:54 |
Dejan | and a definition file per each "command" | 09:54 |
Dejan | so whenever someone does something like select somefield, mycommand(somefield) from sometable | 09:55 |
*** bonro012 has quit IRC | 09:55 | |
Dejan | ingres executes "mycommand" app | 09:55 |
pboro | that would be really, really slow | 09:55 |
Dejan | sure it would | 09:55 |
*** bonro01 has joined #ingres | 09:55 | |
Dejan | but user should be aware of that | 09:55 |
Dejan | this approach surely is not suitable for everything | 09:56 |
DerMeister | Usually users aren't aware of anything until they complain about it :P | 09:56 |
Dejan | but, i do not insert md5 digest of some user every millisecond | 09:56 |
Dejan | or update, or check | 09:57 |
Dejan | ok, depends on the database | 09:57 |
Dejan | naturally | 09:57 |
Dejan | and nature of business | 09:57 |
Dejan | anyways, i need this MD5() very much | 09:58 |
Dejan | beacuse i do not want application to give me MD5 | 09:58 |
Dejan | :) | 09:58 |
Dejan | or SHA1 | 09:59 |
Dejan | does not really matter | 09:59 |
DarylM | cd | 11:36 |
DarylM | oops | 11:36 |
DarylM | ;-) | 11:37 |
toumi01 | just be careful where you type rm -rf ;) | 11:37 |
DarylM | been there, done that | 11:38 |
DarylM | Pretty funny to watch a Unix system eat itself..... | 11:38 |
* DarylM ponders the results of rm -fr / | 11:38 | |
* toumi01 remembers years ago there was a Solaris bug with the user admin GUI - create a user with no home directory so internally home is / - then delete that user with checkbox "delete home directory" checked ;) | 11:41 | |
DarylM | ouch | 11:42 |
*** mull has quit IRC | 12:19 | |
*** mull_ has joined #ingres | 12:19 | |
*** DerMeister has quit IRC | 14:33 | |
*** DarylM has quit IRC | 15:14 | |
*** mull_ has quit IRC | 15:15 | |
*** zxiiro has joined #ingres | 16:26 | |
*** ChanServ sets mode: +v zxiiro | 16:26 | |
*** rossand has quit IRC | 16:49 | |
*** zxiiro has quit IRC | 18:03 | |
*** cytrinox_ has joined #ingres | 19:20 | |
*** cytrinox has quit IRC | 19:36 | |
*** cytrinox_ is now known as cytrinox | 19:36 |
Generated by irclog2html.py 2.7 by Marius Gedminas - find it at mg.pov.lt!