*** Alex| has quit IRC | 00:13 | |
*** zxiiro has joined #ingres | 00:14 | |
*** clach04 has joined #ingres | 00:14 | |
*** pboro has joined #ingres | 00:14 | |
*** card.freenode.net sets mode: +v zxiiro | 00:14 | |
*** ChanServ sets mode: +o withdefault | 00:14 | |
*** Dejan has joined #ingres | 00:24 | |
Dejan | hello everybody | 00:24 |
---|---|---|
Dejan | does anyone here use bzip2 compression for checkpoints ? | 00:25 |
pboro | not anymore, since it's so slow... I prefer z7 | 00:25 |
withdefault | I think that the appliances use bzip2 | 00:25 |
pboro | bzip2 is good for tape backups tho' | 00:26 |
*** Alex| has joined #ingres | 00:34 | |
*** ChanServ sets mode: +o Alex| | 00:34 | |
Dejan | pboro: can you give me your 7z cktmpl.def please ? | 00:36 |
pboro | I'm on a trip right now so not right now, but next week is possible :) | 00:36 |
Dejan | i just need an example... | 00:37 |
Dejan | anyway, thanks | 00:37 |
Dejan | by the time you come back from the trip i will sort this out :D | 00:37 |
pboro | :D | 00:37 |
Dejan | i suppose it would be as easy as with bzip2 | 00:39 |
Dejan | once i get tar with +J flag | 00:39 |
Dejan | what i did on other server was just add "j" flag to all tar commands | 00:40 |
Dejan | and it worked straight away | 00:40 |
*** withdefault sets mode: -o withdefault | 00:43 | |
*** grantc has joined #ingres | 00:45 | |
withdefault | that's pretty much all that's needed - http://ingres.pastebin.com/4TTAufYA | 00:45 |
Dejan | yeah, i suppose so | 00:49 |
Dejan | i hope we will upgrade this server soon | 00:49 |
Dejan | so i can change the compression flag to J | 00:50 |
Dejan | so tar uses lzma | 00:50 |
withdefault | are you using the II_CKTMPL_FILE symbol.tbl symbol to set the checkpoint template or are you modifying the default one? | 00:51 |
Dejan | on my test server i modified cktmpl.def directly | 01:12 |
Dejan | but i made a backup | 01:12 |
Dejan | of the old one | 01:12 |
withdefault | now you know about II_CKTMPL_FILE :) | 01:14 |
Dejan | i know about it from the documentation | 01:16 |
Dejan | but never used it | 01:16 |
Dejan | it is a good idea anyway | 01:17 |
Dejan | makes it easier to switch | 01:17 |
Dejan | withdefault: is it a good idea to specify II_NUM_OF_PROCESSORS=4 if i have 4 cores CPU, or should i set to 1 ? | 01:21 |
withdefault | what platform? | 01:21 |
Dejan | Linux | 01:21 |
Dejan | x86_64 | 01:21 |
withdefault | I think it is used to determine the amount of wait apportioned to semaphores, in mutli-cpu/core on Linux greater than 1, 2 is probably recommended but I'm not 100% | 01:24 |
withdefault | here's what I've found | 01:26 |
withdefault | The value is used in calculations used to detect mutex contention. If we have more processors we can afford to have more active threads spinning on the same mutex before moving to a blocked call. So a value of 4 will move to blocked calls with fewer sessions contending for the same mutex, than if you had a value of 8. It is advisable to set the value to the actual number of processors so that this heuristic will work as expected. Setting a va | 01:26 |
withdefault | lue of 1 will force blocked mutex waits and prevent any active spinning on mutexes. Under some load types (thousands of active sessions) this may provide better performance. | 01:26 |
Dejan | khm, i will leave it at 4 then :) | 01:49 |
Dejan | and see how it goes | 01:49 |
Vroomfondle | what's the advantage of using spinlocks in a database context? | 02:08 |
Dejan | Vroomfondle: probably query optimizer uses hem | 02:10 |
Dejan | them | 02:10 |
Dejan | and similar facilities | 02:10 |
Vroomfondle | I know what one is (purely because it was covered on the Computer Science course I studied) but I don't really understand what the advantage is exactly | 02:12 |
Vroomfondle | wikipedia says "lower OS overheads" so I guess a slight performance increase? | 02:13 |
withdefault | yes, but depending on the number of cpu cycles the spinlock waits can result in higher cpu | 02:14 |
withdefault | higher cpu use | 02:14 |
*** PaulM05 has joined #ingres | 02:18 | |
*** KermitTheFragger has joined #ingres | 02:24 | |
Dejan | ingres takes almost 0% CPU atm | 03:27 |
Dejan | and i have ~80 users connected | 03:28 |
Dejan | ok, i like - 3% | 03:28 |
Dejan | :D | 03:28 |
Dejan | s/like/lie/ | 03:28 |
* Dejan is thinking about starting learning OpenROAD | 03:34 | |
Dejan | is there a quick guide available for it? | 03:34 |
Dejan | i was looking to download it ... failed | 03:34 |
Dejan | where can i download it from? | 03:34 |
PaulM05 | OpenROAD itself is not open-source so you can't download it without a support contract. You should be able to download EMPIRE - which is a community version but it doesn't have all the same functionality. To be honest I'm not sure what the differences are because OpenROAD is not my area | 03:44 |
Vroomfondle | last time I tried, the community download for EMPIRE actually required a copy of OpenROAD | 03:45 |
Vroomfondle | I don't think it covers the internals, just the toolset | 03:45 |
Vroomfondle | (but that was a while ago) | 03:46 |
PaulM05 | as I say I'm not sure what's missing from EMPIRE | 03:46 |
Vroomfondle | out of interest, why's it not open-source? Copyright problems? | 03:49 |
PaulM05 | not sure to be honest | 03:51 |
*** gerhard has joined #ingres | 04:29 | |
*** cthibert has joined #ingres | 04:43 | |
*** zxiiro has quit IRC | 04:52 | |
Dejan | a question... can DB trigger run shell script ? | 04:53 |
Dejan | :D | 04:53 |
PaulM05 | directly no - but you can build mechanisms to do it | 04:54 |
PaulM05 | you can raise a dbevent and have daemon process that listens for them and runs your commands | 04:55 |
Dejan | what i need is a trigger that will update bunch of databases with information from a certain table in the "master database"... | 04:55 |
PaulM05 | you could be sneaky and get the remote command server to do it for you - saves you building a daemon | 04:55 |
PaulM05 | you could use star | 04:56 |
PaulM05 | hmm - maybe not star - doesn't look like you can create a rule on a star database | 05:01 |
PaulM05 | other option would be to have a replicated table and rules on that table | 05:02 |
*** rossand has joined #ingres | 05:22 | |
*** ChanServ sets mode: +o rossand | 05:22 | |
*** mull has quit IRC | 05:26 | |
*** Dejan has quit IRC | 05:52 | |
*** PaulM05 is now known as PaulM05_lunch | 06:09 | |
*** mull has joined #ingres | 06:10 | |
*** zxiiro has joined #ingres | 06:26 | |
*** ChanServ sets mode: +v zxiiro | 06:26 | |
*** atrofast has joined #ingres | 06:42 | |
*** atrofast has quit IRC | 06:45 | |
*** rossand1 has joined #ingres | 06:51 | |
*** cthibert has left #ingres | 06:51 | |
*** rossand has quit IRC | 06:53 | |
*** cthibert has joined #ingres | 06:53 | |
*** gerhard has quit IRC | 06:59 | |
*** PaulM05_lunch is now known as PaulM05 | 07:01 | |
*** Alex| has quit IRC | 07:07 | |
*** atrofast has joined #ingres | 07:08 | |
*** atrofast has quit IRC | 07:10 | |
*** frima01 has joined #ingres | 07:14 | |
*** Mud has joined #ingres | 07:28 | |
*** Mud has quit IRC | 07:43 | |
*** Mud has joined #ingres | 08:29 | |
*** frima01 has quit IRC | 08:31 | |
*** Mud has quit IRC | 08:52 | |
*** grantc has quit IRC | 08:56 | |
*** Mud has joined #ingres | 08:58 | |
*** DarylM-mini9 has joined #ingres | 09:07 | |
*** DarylM-mini9 has quit IRC | 09:09 | |
*** DarylM-Mini9 has joined #ingres | 09:32 | |
*** Alex| has joined #ingres | 09:50 | |
*** ChanServ sets mode: +o Alex| | 09:50 | |
*** Alex| has quit IRC | 10:35 | |
*** DarylM-Mini9 has quit IRC | 10:46 | |
*** KermitTheFragger has quit IRC | 11:07 | |
*** PaulM05 has quit IRC | 11:41 | |
*** mull has quit IRC | 12:24 | |
*** ccsidiot has joined #ingres | 12:25 | |
*** rossand1 is now known as rossand | 12:33 | |
*** atrofast has joined #ingres | 12:41 | |
*** Mud has quit IRC | 12:53 | |
*** Mud has joined #ingres | 12:58 | |
*** ccsidiot has quit IRC | 13:02 | |
*** rossand has quit IRC | 13:11 | |
*** atrofast has quit IRC | 13:26 | |
*** cthibert has left #ingres | 13:31 | |
*** rossand has joined #ingres | 14:34 | |
*** ChanServ sets mode: +o rossand | 14:34 | |
*** ccsidiot has joined #ingres | 14:36 | |
*** zxiiro has quit IRC | 14:50 | |
*** rossand has quit IRC | 15:00 | |
*** rossand1 has joined #ingres | 15:00 | |
*** rossand1 has quit IRC | 15:00 | |
*** withdefault has quit IRC | 15:16 | |
*** Mud has quit IRC | 15:27 | |
*** zxiiro has joined #ingres | 15:27 | |
*** ChanServ sets mode: +v zxiiro | 15:27 | |
*** bonro01 has quit IRC | 15:32 | |
*** bonro01 has joined #ingres | 15:36 | |
*** bonro01 has quit IRC | 15:46 | |
*** bonro01 has joined #ingres | 15:50 | |
*** bonro01 has quit IRC | 15:56 | |
*** bonro01 has joined #ingres | 15:59 | |
*** FrankW has quit IRC | 16:07 | |
*** FrankW- has joined #ingres | 16:07 | |
*** zxiiro_ has joined #ingres | 16:12 | |
*** zxiiro has quit IRC | 16:12 | |
*** zxiiro_ is now known as zxiiro | 16:12 | |
*** ChanServ sets mode: +v zxiiro | 16:12 | |
*** zxiiro has quit IRC | 16:19 | |
*** zxiiro has joined #ingres | 16:22 | |
*** ChanServ sets mode: +v zxiiro | 16:22 | |
*** ccsidiot has quit IRC | 17:53 | |
*** rossand has joined #ingres | 19:03 | |
*** ChanServ sets mode: +o rossand | 19:03 | |
*** cytrinox_ has joined #ingres | 20:20 | |
*** cytrinox has quit IRC | 20:23 | |
*** cytrinox_ is now known as cytrinox | 20:23 | |
*** rossand has quit IRC | 21:28 | |
*** Alex| has joined #ingres | 23:48 | |
*** ChanServ sets mode: +o Alex| | 23:48 |
Generated by irclog2html.py 2.7 by Marius Gedminas - find it at mg.pov.lt!