grantc | morning | 00:06 |
---|---|---|
*** PaulM05 has joined #ingres | 01:12 | |
*** KermitTheFragger has joined #ingres | 01:16 | |
Deyan | short question - if i do a checkpoint on server1, scp it to server2, can i recreate the database by using only that, single checkpoint ? | 02:02 |
Deyan | (on server2) | 02:03 |
Deyan | (on server2 i regularly delete that database, it is a test server) | 02:04 |
Deyan | i know i can use unloaddb, etc, but i was just curious | 02:05 |
PaulM05 | you can but only if you have the same locations, in the same order with the same physical paths | 02:05 |
Deyan | yep, they are on identical locations | 02:06 |
Deyan | PaulM05: thanks you | 02:07 |
Deyan | s/thanks/thank/ | 02:08 |
PaulM05 | there's a Support knowledgebase document on it - just checking if there's the equivalent on the wiki | 02:08 |
PaulM05 | can't find it :( | 02:16 |
Deyan | it is fine, i appreciate the effort anyway | 02:20 |
PaulM05 | ok - sorry about that | 02:20 |
Deyan | i need to practice database recovery... :) | 02:20 |
Deyan | problem is, i do not have time to do that :) | 02:21 |
Deyan | i am a ful-time software developer, i am not supposed to be DBA | 02:21 |
PaulM05 | :) | 02:22 |
Deyan | i am thinking to refine this backup system to generate HTML report so our system managers can see if everything went well... | 02:23 |
Deyan | and an idea just came to my head, that is not Ingres related... | 02:23 |
Deyan | but kind of intersting for generating HTML... | 02:23 |
Deyan | i am thinking to write a set of BASH functions | 02:23 |
Deyan | that could give me ability to do something like: | 02:24 |
Deyan | #!/bin/sh | 02:24 |
Deyan | html | 02:24 |
Deyan | title "report page" | 02:24 |
Deyan | body | 02:24 |
Deyan | para "some text" | 02:24 |
Deyan | body_ | 02:24 |
Deyan | html_ | 02:24 |
Deyan | sorry for spam | 02:25 |
Deyan | this small peace of valid BASH code could actually be enough to generate a simple HTML | 02:25 |
Deyan | khm, i kinda like it | 02:25 |
PaulM05 | sounds good - brb | 02:26 |
* grantc wonders why Deyan does not use Perl/PHP/etc.. | 02:26 | |
Deyan | grantc: i would if i do not want this not to introduce any unnecessary dependencies... | 02:27 |
grantc | the way I monitor the backups is to receive an email of the output if $? != 0 | 02:28 |
Deyan | and coding this above is piece of cake | 02:28 |
Vroomfondle | does Ingres have an equivalent of "if not exists" when creating a table? Or do I *have* to check iitables to see if the table already exists? | 03:20 |
grantc | Vroomfondle, you have to check | 03:21 |
PaulM05 | there's no "CREATE OR REPLACE" at the moment - it came up at the last code sprint but it's not been implemented yet | 03:21 |
Vroomfondle | ah, k | 03:21 |
*** cthibert has joined #ingres | 03:53 | |
Deyan | PaulM05: not even in the latest build ? | 04:19 |
PaulM05 | unless it's very recent | 04:20 |
PaulM05 | I haven't seen a change that implements it | 04:20 |
Deyan | i have a weir problem here ... two servers, 100% same data, and query produces different output | 04:38 |
Deyan | server1 gives 4 rows, server2 gives 2 rows | 04:38 |
PaulM05 | what's the difference between the servers? | 04:44 |
Deyan | ok, help table helped me to find the difference | 04:51 |
Deyan | server1 table had indices | 04:51 |
Deyan | server2 did not | 04:51 |
Deyan | funny thing is, server2 produces correct result | 04:51 |
Deyan | (without indices) | 04:51 |
PaulM05 | could be that the index are corrupt - you could try re-creating them - or it could just be a bug :) | 04:52 |
*** atrofast has joined #ingres | 04:57 | |
Deyan | i am thinking about doing it | 05:01 |
Deyan | to recreate index... | 05:01 |
Deyan | PaulM05: there is a difference in config, that i just realised | 05:01 |
Deyan | page size is different | 05:01 |
Deyan | can that cause difference ? | 05:02 |
Deyan | IMHO it should not... | 05:02 |
PaulM05 | you're right | 05:02 |
PaulM05 | but first we need to find out the difference that's causing the behaviour then we can pinpoint the bug (it does sound like a bug) | 05:03 |
Deyan | yeah, i agree | 05:03 |
Deyan | let's first find out if i will get the same problem after i recreate indices | 05:03 |
PaulM05 | yep - good plan | 05:04 |
*** atrofast has quit IRC | 05:07 | |
PaulM05 | excuse me I'm off to lunch now | 05:07 |
*** PaulM05 is now known as PaulM05_lunch | 05:07 | |
*** atrofast has joined #ingres | 05:12 | |
*** rossand has joined #ingres | 05:39 | |
*** ChanServ sets mode: +o rossand | 05:39 | |
*** PaulM05_lunch is now known as PaulM05 | 05:46 | |
*** gerhard has joined #ingres | 05:53 | |
gerhard | Hi all | 05:53 |
PaulM05 | hi gerhard | 05:54 |
gerhard | I suppose the "...user does not have privileges in config.dat..." message is well known to all who have to sometimes add / change vnode entries on Windows clients...? | 06:06 |
gerhard | At time of installation, Ingres installer (for newer 9.x.x releases) will add "...privileges..." lines for all users in admin group to config.dat, right? | 06:12 |
PaulM05 | tbh I don't know | 06:13 |
PaulM05 | are you trying to create global vnodes as a non-Admin user? | 06:14 |
gerhard | I think it is that way. | 06:15 |
gerhard | the problem: an admin user is added to a Windows (client) machine *after* Ingres installation | 06:18 |
gerhard | ...and that user - despite the fact being admin - is not capable of adding / changing vnode entries with netutil or VDBA | 06:19 |
PaulM05 | yes because it's the config.dat privileges that count | 06:20 |
gerhard | right, two workarounds: | 06:37 |
gerhard | 1. log off / log in as a user that is in the config.dat | 06:38 |
gerhard | 2. "notepad config.dat", find existing privileges line, make a copy for desired user | 06:39 |
gerhard | and a variant of 2. | 06:39 |
gerhard | iisetres ii.%computername%.privileges.user.%username% SERVER_CONTROL,NET_ADMIN,MONITOR,TRUSTED | 06:40 |
*** DarylM has joined #ingres | 06:40 | |
gerhard | of course, having to ingstop / ingstart so Ingres reads new config.dat... | 06:41 |
gerhard | ...which will introduce a new problem because ingstop might also throw the "user does not have privileges messages" | 06:42 |
PaulM05 | if it's configured to start/stop as a service you can get around that | 06:42 |
gerhard | OK, you're right, "net stop ingres_database_ii" is an alternative... | 06:43 |
*** cthibert has left #ingres | 06:44 | |
gerhard | but nevertheless, my customers complain about Ingres being not very "Windows-compliant" here and to be honest, they're right... | 06:45 |
*** cthibert has joined #ingres | 06:48 | |
gerhard | why not throw away all that privileges lines stuff and make it like any Windows admin would expect it....? | 06:49 |
PaulM05 | because it has to work on non-Windows platforms as well | 06:50 |
gerhard | --> user is Windows-admin --> then he is also Ingres-admin | 06:51 |
PaulM05 | that would be a nice feature - but I wouldn't want to "throw away" what's there already | 06:51 |
gerhard | OK, that was a little bit harsh (-: | 06:52 |
gerhard | don't get me wrong: I like Ingres' consistent look-and-feel on both Win and *nix systems | 06:53 |
gerhard | but why not make a distinction where it really *does* make sense. | 06:55 |
PaulM05 | like I said it would be a nice feature - are you offering to code it? ;) | 06:58 |
gerhard | There also has been a *VDBA* around for years only for Windows machines and I think that was not a big problem... | 06:59 |
gerhard | ...because people consider it normal to work the GUI way on Windows and the CLI way on Linux. | 07:00 |
PaulM05 | we did have VDBA on Linux and Unix for a while - but not since we went Open Source as it used a proprietary library | 07:01 |
gerhard | That means 2.5 / 2.6 on Linux had a VDBA? Really? | 07:02 |
PaulM05 | 2.6 only I'm not sure it was in 2.5 | 07:03 |
gerhard | I'm quite curious about the new Java-based admin stuff. Are there any (community) Ingres releases around that already ship this? | 07:06 |
PaulM05 | not yet | 07:06 |
PaulM05 | I don't think there's enough there yet to ship | 07:07 |
PaulM05 | but I got the impression that they'd be looking to put something out to the community | 07:08 |
PaulM05 | when I don't know | 07:09 |
*** atrofast has quit IRC | 07:12 | |
gerhard | Another issue / question concerning Ingres on Windows: we have been using Ingres on Windows since Ingres II 2.0 | 07:14 |
gerhard | and most of the time, it was mandatory to have an operating system user "ingres" on the machine. | 07:15 |
grantc | gerhard, that's no longer the case since 9.1 (perhaps 9.0) | 07:16 |
gerhard | yep | 07:16 |
grantc | regarding your windows user issue - are they configuring installation passwords or a hard coded user/passwd for the server? | 07:17 |
gerhard | hard coded user / pass | 07:18 |
grantc | then I propose two solutions/alternative mechanisms | 07:19 |
gerhard | for our (future) Windows server machines we are thinking about installing without an os user ingres... | 07:19 |
grantc | either have the application use a connection string of the form "@server,tcp_ip,listenaddress[user,passwd]::db" | 07:19 |
gerhard | ...but are unsure about the consequences. | 07:19 |
grantc | or remotely configure netuil from the server | 07:20 |
gerhard | Yes, I know that connection string stuff, have successfully tried it for sql / isql | 07:21 |
grantc | it should work for all the interfaces afaik | 07:22 |
gerhard | problem: we have some (Crystal Report) applications that need ODBC-System-DSNs and for the DSN you first need a vnode entry | 07:22 |
grantc | then your only and perhaps best option is to configure each PC remotely using a script | 07:23 |
grantc | in fact once you have written the script it's quicker to remotely do it rather than go to each PC. | 07:24 |
gerhard | or could I fiddle that @server.... stuff into an ODBC DSN...? Don't think it is possible the GUI way but maybe directly in the registry or with odbcconf command...? | 07:24 |
grantc | I did it for a CA world conference | 07:24 |
grantc | I don't know how it could be done for a DSN | 07:25 |
*** Mud has joined #ingres | 07:25 | |
grantc | It's worth raising an issue for | 07:25 |
*** mull has joined #ingres | 07:25 | |
gerhard | You mean kicking off some "netutil -f answerfile" action remotely by using tools like psexec, right? | 07:26 |
gerhard | I have sometimes worked around the config.dat issue by right-clicking on the DOS-Box shortcut and do a "run-as" | 07:27 |
gerhard | I'll test a little bit with the DSN stuff and give feedback maybe in a few days... | 07:27 |
grantc | something like that although you'll need to have a vnode defined for each client PC IIRC on the PC/server doing the remote configuring | 07:28 |
gerhard | back to the the topic running Ingres as ingres vs. running Ingres as built-in system account on Windows servers... | 07:29 |
gerhard | how could I manage - without an OS user ingres - that the owner for each table in DB is still ingres? | 07:30 |
grantc | the owner of each table will be whoever created it | 07:32 |
gerhard | It was always convention for us that the DB owner is ingres and the owner of all DB objects is also ingres | 07:32 |
grantc | without an OS user I'm not sure how you'll authenticate as ingres | 07:33 |
gerhard | So that means if Ingres-DBMS installed to run as built-in system account (also sometimes titled LocalSystem)... | 07:33 |
gerhard | ...and being logged in as Administrator... | 07:33 |
grantc | you can now install as anyuser provided that user has the necessary rights | 07:34 |
gerhard | I could do createdb... / isql ... / create table... | 07:34 |
grantc | so if you install ingres as admin then admin is the user you will use | 07:34 |
gerhard | ...and everything would be owned by Administrator, right? | 07:34 |
grantc | Admin is created as a default user IIRC | 07:34 |
grantc | yup | 07:34 |
*** Deyan is now known as Dejan | 07:35 | |
gerhard | could I run something like "create user ingres identified by Ingres..." against iidbdb even if ingres is not an os user on the server? | 07:36 |
grantc | you can create users with a dbms password | 07:39 |
grantc | however the dbms password is in addition to an OS password, unless you either enable null authentication or use installation passwords | 07:39 |
*** Mud has quit IRC | 07:44 | |
gerhard | null authentication: yes, I know that dirty little trick (-: | 07:44 |
gerhard | create user with a dbms password - how would I do that? | 07:45 |
*** Mud has joined #ingres | 07:50 | |
grantc | create user username with password = 'secret' | 07:57 |
*** Alex| has quit IRC | 07:57 | |
grantc | however every app has to specify that password on connection using the dbms password flag or parameter | 07:58 |
grantc | not sure how it would work using an ODBC DSN / without additional code | 07:58 |
gerhard | how would that look like for usage of isql? I have used the -uingres Flag in the past, but only for systems where ingres existed as OS user. | 07:59 |
grantc | i think it's -P | 07:59 |
gerhard | how would I grant DBA privs to user ingres? | 08:00 |
gerhard | will test that out on a VM | 08:00 |
grantc | either using accessdb / vdba when editing a user or using the relevant privileges with "create user" | 08:01 |
*** PaulM05 has quit IRC | 08:01 | |
grantc | http://docs.ingres.com/Ingres/9.2/SQL%20Reference%20Guide/createuser.htm | 08:01 |
*** gerhard has quit IRC | 08:05 | |
Dejan | how can i enable journaling of imadb ? | 08:09 |
grantc | ckpdb +j imadb | 08:13 |
grantc | the ima tables we supply are not journalable | 08:13 |
*** zxiiro has joined #ingres | 08:42 | |
*** ChanServ sets mode: +v zxiiro | 08:42 | |
*** Mud has quit IRC | 08:47 | |
*** grantc has quit IRC | 08:54 | |
*** Mud has joined #ingres | 09:08 | |
FrankW | I have a table field for which the C API descriptor looks like the following when I do a "select * from tpoly": | 09:34 |
FrankW | {ds_dataType = 61, ds_nullable = 1, ds_length = 65534, ds_precision = -1, ds_scale = 0, ds_columnType = 0, ds_columnName = 0x607cc0 "shape"} | 09:34 |
FrankW | I do not see a datatype corresponding to 61 in the iiapi set of IIAPI_*_TYPE values. How should I interprete it? | 09:35 |
cthibert | FrankW type 61 should be DB_POLY_TYPE and 62 is DB_MPOLY_TYPE. I | 09:52 |
FrankW | where would I find these declared? | 09:52 |
cthibert | I'm looking further to determine why it's not showing up right. It's declared in common/hdr/hdr/iicommon.h | 09:52 |
cthibert | And types are defined in common/adf/adg/adgdttab.roc | 09:53 |
cthibert | Probably haven't made it to the API definitions yet. I'll look. | 09:53 |
FrankW | iicommon.h does not appear to be installed in the "files" are of the database. | 09:53 |
FrankW | warmerda@gdal64[71]% pwd | 09:54 |
FrankW | warmerda@gdal64[72]% find . -name iicommon.h -print | 09:54 |
FrankW | wrk/home/warmerda/ingres/server/install/II/ingres was my directory. | 09:54 |
cthibert | It's in the code. src/common | 09:54 |
FrankW | OGR's configure expects to be pointed at an installed database for include files. | 09:55 |
cthibert | Yup. Understood. I'm looking for where those definitions should be in our code to ensure they get to the right place. | 09:55 |
FrankW | my attempt to read back a "select * from tpoly" in OGR is failing because as soon as I try to do IIapi_getColumns() to fetch data for the type 61 column, I get IIAPI_ST_NODATA. | 09:56 |
FrankW | which I interprete as the end of the resultset. | 09:56 |
FrankW | IIAPI_ST_NO_DATA I should say. | 09:56 |
FrankW | I even tried forcing a continue, but I get IIAPI_ST_NO_DATA for all subsequent requests. | 09:57 |
FrankW | For now I've just disabled this one test in my test script, but in the long term I hope we can address it. | 09:57 |
cthibert | Absolutely. The cause appears to be not propagating the types to the API headers. I'll be updating that soon. | 09:58 |
FrankW | I have created a gdal ticket for tracking the issue. | 10:03 |
cthibert | FrankW - Another note to this. The data held in the table isn't pure WKB. It's a stream of data that includes the WKB. So, a select * won't return 'usuable' data from an API view. The asBinary/asText calls parse out the appropriate data. | 10:14 |
cthibert | The data types should still be reflected in the API, but just wanted you to be aware that it's not a 'pure' WKB stream that will be retrieved via that call. | 10:16 |
*** KermitTheFragger has quit IRC | 10:17 | |
FrankW | cthibert: understood | 10:18 |
*** Alex| has joined #ingres | 10:19 | |
*** ChanServ sets mode: +o Alex| | 10:19 | |
Dejan | can someone explain what do i need to make a copy of a database by using .ckp file only? | 10:27 |
Dejan | i have server1 and server2 | 10:27 |
Dejan | i moved ckp file from server1 to server2 | 10:27 |
Dejan | (they have the same locations) | 10:27 |
Dejan | what else should i copy in order to use rollforwarddb to create copy of the database on server2 via rollforwarddb ? | 10:28 |
Dejan | i usually do unloaddb | 10:28 |
Dejan | for this | 10:29 |
Dejan | but for some reasons i need this approach now | 10:29 |
Dejan | PaulM told me it is possible | 10:29 |
Dejan | but i am doing something wrong :) | 10:29 |
*** DerMeister has joined #ingres | 10:42 | |
*** atrofast has joined #ingres | 10:46 | |
Dejan | \q | 10:55 |
*** Dejan has quit IRC | 11:05 | |
*** Mud has quit IRC | 11:44 | |
*** Mud has joined #ingres | 11:50 | |
*** rossand1 has joined #ingres | 11:50 | |
*** rossand has quit IRC | 11:52 | |
*** DerMeister has quit IRC | 12:34 | |
*** Alex| has quit IRC | 12:55 | |
*** cthibert has left #ingres | 13:57 | |
*** zxiiro has quit IRC | 14:00 | |
*** zxiiro has joined #ingres | 15:09 | |
*** ChanServ sets mode: +v zxiiro | 15:09 | |
*** Mud has quit IRC | 15:46 | |
*** Mud has joined #ingres | 15:50 | |
*** mull has quit IRC | 15:57 | |
*** rossand1 has quit IRC | 17:11 | |
*** cytrinox_ has joined #ingres | 19:21 | |
*** cytrinox has quit IRC | 19:23 | |
*** cytrinox_ is now known as cytrinox | 19:23 | |
*** Mud has quit IRC | 19:44 | |
*** Mud has joined #ingres | 19:50 | |
*** zxiiro has quit IRC | 20:06 | |
*** Adross has joined #ingres | 21:17 | |
Adross | I am trying to design an ingres query that will select the primary key of a given table, the name of each of the tables that the given table references, and the primary key of each of those tables. In mysql, I would use information_schema.key_column_usage table, but I'm not sure which table to use on ingres. The closest is iiref_constraints | 21:19 |
Adross | so, basically, is there a way to output the primary key of a given table in ingres | 21:20 |
*** Alex| has joined #ingres | 22:38 | |
*** ChanServ sets mode: +o Alex| | 22:38 | |
*** Adross has quit IRC | 22:52 | |
*** Mud has quit IRC | 23:03 | |
*** Mud has joined #ingres | 23:36 | |
*** Mud has quit IRC | 23:45 | |
*** Mud has joined #ingres | 23:50 |
Generated by irclog2html.py 2.7 by Marius Gedminas - find it at mg.pov.lt!