*** grantc has joined #ingres | 00:06 | |
*** ChanServ sets mode: +o grantc | 00:06 | |
*** Deyan has joined #ingres | 01:15 | |
Deyan | good morning | 01:15 |
---|---|---|
magnusg| | God morning Deyan | 01:39 |
*** dyki has quit IRC | 02:44 | |
pboro | oh my... the price of HVR is... stunning | 03:04 |
pboro | HVR for two databases costs more in a year than our Ingres subscriptions in five years | 03:05 |
pboro | well, at least it's an easy decision :) | 03:05 |
*** dyki has joined #ingres | 03:21 | |
pboro | If I really, really, really needed hot-standby option, I guess any other DBMS would be cheaper than Ingres :( Even Oracle. | 03:25 |
magnusg| | pboro, you can use incremental rollforwarddb to get a stand by, if not hot so at least pretty warm. that's in 9.2 though. Many of our customers use a failover cluster to get a stand by. | 03:50 |
pboro | magnusg|, yeah checkpoint every 30 minutes is enough considering the price of HVR :( | 03:51 |
magnusg| | You don't need to take checkpoints often - you need to copy the journal files. | 03:52 |
pboro | it's not journalled | 03:52 |
pboro | but yeah, that's true | 03:53 |
pboro | still stunned by the prices, damn... and most other dbmses even support hot-standby (one-way log based replication) as an integrated feature | 03:54 |
magnusg| | pboro, I think there has been discussions in that direction with Ingres too - as a way of improving the Replicator, but I think the decision has been not to invest in the Replicator. | 03:58 |
pboro | the Replicator is trigger based and pretty ugly, it should be pretty easy to create a new module that would tap into logging system, read the log and send the data over to another server which would then apply the log operations into the database | 03:59 |
magnusg| | Even if Oracle has bought Golden Gate it could be worth while checking what they offer too. | 04:00 |
pboro | yup | 04:00 |
magnusg| | Yes the current Replicator can only handle small number of transactions, or at least not high numbers. | 04:01 |
pboro | I believe that couple skilled Ingres developers could create a simple hot-standby feature in Ingres under a month :) | 04:02 |
pboro | I could be wrong, but just based what I have been digging the code and talking around. I would even try it myself if I had the time :) | 04:03 |
thiagomz | guys, what is XROW lock ? | 04:34 |
pboro | exclusive row lock? | 04:34 |
magnusg| | Hi all, I had a question from a customer regarding Ingres 9.2 build numbers. The current bulld (enterprise version) is 143. They was asking when to expect a new. I suppose that would be a service pack or Ingres 9.3 - Is that correct? | 04:52 |
grantc | magnusg|, service packs would be 9.2.1, 9.2.2 etc... | 04:54 |
grantc | 9.3 and 10.0 are major releases | 04:54 |
grantc | as to if there will be a service pack for 9.2, you would have to contact support to get an answer | 04:57 |
magnusg| | grantc, they asked about the build number (143). Is it correct that it is only changed with service packs or major releases? | 05:04 |
grantc | it used to be the case that build numbers went from 99 to 0 and that 0 = final release | 05:04 |
grantc | now they go from 100 upwards... | 05:04 |
grantc | let me check to see whether 9.1.1 had build numbers > 1 | 05:05 |
grantc | 100 even | 05:05 |
grantc | SPs follow the same rules - 100 = first build, then incrementing | 05:06 |
magnusg| | But when it's GA the number is kept until a new SP I suppose. Patches have the same build number as the GA version. If feels like build are used pretty much like the old GEN LEVLS | 05:09 |
magnusg| | Enterprise versions that is. Community version might differ. | 05:09 |
grantc | yup so 9.1.1 (...../103) is newer than 9.1.0 (..../123) | 05:10 |
grantc | the genlevel was CA's released to QA date IIRC | 05:11 |
grantc | community releases don't get patches but apart from that they follow the same naming scheme (more or less) | 05:12 |
* grantc wanders off to get some fruit. | 05:13 | |
* grantc is away: wanders off | 05:13 | |
*** DarylM has joined #ingres | 05:40 | |
pboro | hehe, some good comments in the archiver code... "I've left this code in because it compiles. However, I don't think it works and I don't know what we should do with it." | 05:57 |
*** rossand has joined #ingres | 05:59 | |
*** ChanServ sets mode: +o rossand | 05:59 | |
pboro | Hm hm... it was rather easy to copy and modify the archiver code to read the log and output some records on screen... but since archiver and logging system talk to each other, a log-based replicator would also need some modifications to the logging system to make logging system aware of it | 06:01 |
*** mull has joined #ingres | 06:22 | |
*** magnusg| has quit IRC | 07:34 | |
*** dyki has quit IRC | 07:47 | |
*** mull has quit IRC | 08:03 | |
*** mull has joined #ingres | 08:03 | |
*** grantc has quit IRC | 09:14 | |
*** Deyan has quit IRC | 10:05 | |
*** mull has quit IRC | 11:07 | |
*** mull_ has joined #ingres | 11:07 | |
*** rossand has quit IRC | 11:16 | |
*** DerMeister has joined #ingres | 12:05 | |
*** thiagomz has quit IRC | 12:41 | |
*** mull_ has quit IRC | 13:00 | |
*** DerMeister has quit IRC | 14:52 | |
*** DarylM has quit IRC | 14:53 | |
*** zxiiro_ has joined #ingres | 15:14 | |
*** zxiiro has quit IRC | 15:14 | |
*** zxiiro_ is now known as zxiiro | 15:14 | |
*** ChanServ sets mode: +v zxiiro | 15:14 |
Generated by irclog2html.py 2.7 by Marius Gedminas - find it at mg.pov.lt!