*** dyki has joined #ingres | 00:25 | |
*** Deyan has joined #ingres | 01:58 | |
*** Gerhard has joined #ingres | 02:34 | |
*** Gerhard has quit IRC | 03:28 | |
*** atrofast has joined #ingres | 04:14 | |
grantc | morning all | 05:06 |
---|---|---|
Alex| | had a long night? ;) | 05:06 |
pboro | hehe :) | 05:06 |
grantc | Deyan, could you try http://ingres.pastebin.com/f645b750b (a patch) against the _current SVN_ source for the PHP driver | 05:06 |
grantc | it might fix your close problem | 05:06 |
grantc | Alex|, hmmm | 05:07 |
*** rossand has joined #ingres | 05:12 | |
*** ChanServ sets mode: +o rossand | 05:12 | |
*** rossand has quit IRC | 05:19 | |
Deyan | grantc: i will | 05:30 |
Deyan | ty | 05:30 |
grantc | there is no guarantee it will fix your problem but i found an issue in ingres_query() related to non-select statements | 05:30 |
*** rossand has joined #ingres | 05:32 | |
*** ChanServ sets mode: +o rossand | 05:32 | |
grantc | morning rossand | 05:38 |
rossand | Good afternoon grantc, everyone. | 05:38 |
Deyan | grantc: first run of my script crashed the server | 05:45 |
Deyan | :D | 05:45 |
grantc | not my problem :P | 05:45 |
grantc | paste the errlog.log into the pastebin | 05:46 |
*** Alex| is now known as Alex|off | 05:57 | |
thiagomz | masters... what is the diference between a Consistent Point and a Write Behind ? | 06:00 |
magnusg| | thiagomz, I don't know if I qualify to be a master but a consistency point is a the name of an activity in Ingres where the data caches in memory are flushed to disk, the archiver copies information from transaction log to journal file, the transaction log i updated with some information about this and maybe a few other things are happening too. | 06:33 |
magnusg| | The purpose is to get a point in time where the system is as synchronized on disk as possible, | 06:33 |
magnusg| | Write behind threads are used by a protocol called write ahead (IIRC) which together with fast commit servers means that changes in data is primarily updated in memory caches and in the transaction log and written to the database at a later time. THis being either when the memory cache is getting filled or at a consistency point. | 06:36 |
*** mull has joined #ingres | 06:37 | |
*** atrofast has quit IRC | 06:37 | |
magnusg| | Since I am writing I can also say hello to everybody and say that I've had quite a good time today and yesterday creating a shell script which produces table name, table owner and data file name on tables which has not been updated for a given period of time or the opposite of that. | 06:42 |
*** rossand has quit IRC | 06:59 | |
*** FrankW has joined #ingres | 07:25 | |
*** Deyan_ has joined #ingres | 07:49 | |
Deyan_ | grantc: are you there? | 07:49 |
*** Deyan_ has quit IRC | 07:50 | |
grantc | sort of | 07:50 |
Deyan | i made a paste | 07:50 |
Deyan | i ran script few times | 07:50 |
grantc | i might have another diff for you to try :D | 07:50 |
Deyan | and each time i got some serious errors | 07:50 |
Deyan | awesome | 07:50 |
grantc | again i don't know it will fix your problem but i certainly fixes mine ... | 07:51 |
Deyan | sure | 07:51 |
Deyan | i will be glad to test it | 07:51 |
grantc | svn revert then apply http://ingres.pastebin.com/f5efb65b6 | 07:52 |
grantc | regarding your SEGV - unfortunately as before we will need a test case to resolve it | 07:53 |
Deyan | i frankly do not know how to make a test case :D | 07:54 |
grantc | but it looks like something bad happens in the semaphore code | 07:54 |
Deyan | because i have no clue where the problem lies... | 07:54 |
Deyan | :( | 07:54 |
grantc | toumi01, are you about? | 07:54 |
Deyan | grantc: i can just describe what script does... | 07:55 |
toumi01 | yup | 07:55 |
Deyan | there are few important things to note about this script | 07:55 |
grantc | do you have 5 secs to take a look at an errlog segment | 07:55 |
toumi01 | fire away | 07:55 |
Deyan | 1) few of us may run it at the same time | 07:55 |
grantc | Deyan, what version of ingres is this? | 07:56 |
Deyan | 2) each process connects to a different (test database), which is the target database | 07:56 |
Deyan | 9.3 | 07:56 |
grantc | and the build number (at the end of the brackets)? | 07:56 |
Deyan | Ingres Linux Version II 9.3.0 (a64.lnx/140)NPTL | 07:56 |
Deyan | 3) each of scripts connects also to at least 2 other databases | 07:59 |
Deyan | these 2 databases are common for all processes | 07:59 |
grantc | all the dbs are in the same instance? | 08:01 |
Deyan | yes | 08:03 |
Deyan | later they will be remote | 08:03 |
Deyan | this is just for testing | 08:03 |
Deyan | ok grant | 08:04 |
Deyan | going to apply your new patch | 08:04 |
Deyan | brb | 08:04 |
grantc | k | 08:04 |
*** magnusg| has quit IRC | 08:07 | |
*** toumi01 has left #ingres | 08:16 | |
Deyan | grantc: now it did not cause server-errors | 08:20 |
Deyan | so the new patch is better | 08:20 |
Deyan | IMHO | 08:20 |
Deyan | :D | 08:20 |
grantc | sweet... | 08:20 |
grantc | does the script work now? | 08:21 |
Deyan | i just ran it as single process | 08:21 |
Deyan | did not try to run two in parallel | 08:21 |
Deyan | it is still working | 08:21 |
Deyan | but i can see everything goes fine | 08:22 |
Deyan | previous patch caused crash pretty fast | 08:22 |
*** DerMeister has joined #ingres | 08:26 | |
Deyan | grantc: i am getting bunch of strange warnings | 08:38 |
Deyan | PHP Warning: ingres_query(): Error: 7 - Unable to close non-result resource statement in /home/ingres/dejan/athena/migrator/inc/sys_codes.php on line 9 | 08:38 |
Deyan | PHP Stack trace: | 08:38 |
Deyan | this line contains ingres_query() statement | 08:38 |
Deyan | that is why i find it strange | 08:38 |
Deyan | i get similar warnings in other ingres_query() lines | 08:39 |
grantc | i am passing a bad handle in the code - that's what 7 is | 08:42 |
*** rossand has joined #ingres | 08:42 | |
*** ChanServ sets mode: +o rossand | 08:42 | |
grantc | since you cannot have multiple active deletes at the same time it's trying to clean up from the last one | 08:43 |
Deyan | script is never deleting | 09:05 |
Deyan | if that is what you mean | 09:05 |
Deyan | it is just reading from source(s), and inserting read data into the target database | 09:06 |
Deyan | it never deletes anything | 09:06 |
Deyan | ok, i am going to try parallel execution now | 09:08 |
grantc | deletes/inserts/updates | 09:10 |
grantc | DML | 09:10 |
Deyan | does that mean that some rows may have not been inserted? | 09:15 |
grantc | i don't believe that should be a problem but you should check | 09:16 |
grantc | btw can you paste the table definition for dir_dat_companies | 09:16 |
Deyan | it is not just that table | 09:17 |
grantc | i am more concerned with the server crashing you were having | 09:18 |
Deyan | aham | 09:18 |
Deyan | ok | 09:18 |
Deyan | hold on | 09:18 |
*** toumi01 has joined #ingres | 09:39 | |
dyki | its nice IBM server so it is even more strange :) | 09:39 |
dyki | it has no_crash set as default :) | 09:40 |
Deyan | ROFL | 09:40 |
*** Deyan has quit IRC | 09:51 | |
*** Alex|off is now known as Alex| | 09:57 | |
*** dyki has quit IRC | 10:07 | |
*** Alex| is now known as Alex|off | 10:14 | |
*** grantc_ has joined #ingres | 10:45 | |
*** ChanServ sets mode: +o grantc_ | 10:45 | |
*** grantc has quit IRC | 10:46 | |
*** grantc_ has quit IRC | 10:53 | |
FrankW | rossand: I'm seeing: Link /wrk/home/warmerda/ingres/server/build/utility/unimapcompile | 11:22 |
FrankW | wrk/home/warmerda/ingres/server/build/lib/libq.1.so: undefined reference to `GEOSWKTReader_destroy_r' | 11:22 |
FrankW | It seems somewhat likely that I need a 32bit geos build to link 32bit apps against - is that likely true? | 11:22 |
FrankW | I'm starting to think I should just abandon doing this on 64bit linux. | 11:22 |
rossand | Hi FrankW No question 32 bit is easier. We'd like to sort out any issues on 64 bit if you'd like to fight through them with us? I can ask Chuck to give you a hand - he did the work to integrate GEOS. | 11:24 |
FrankW | OK, I'm game to keep pushing. | 11:25 |
FrankW | I did notice that geos had been added to the link list in Jamefs.int_lnx, but not in Jamdefs.i64_lnx so I added it there. But I only have a 64bit geos on my system. | 11:25 |
rossand | OK, thanks. Chuck's going to be online momentarily. | 11:26 |
rossand | Worst case, I have an extra desktop system we could use if we need to set up a place for you to work. | 11:27 |
*** cthibert has joined #ingres | 11:27 | |
rossand | There's the man. :-) | 11:28 |
cthibert | Hello | 11:28 |
FrankW | Hi Chuck | 11:28 |
FrankW | I'm trying to build the geospatial branch on 64bit linux and getting problems when linking unimapcompile about GEOS functions being missing. | 11:28 |
FrankW | I suspect it is because I lack a 32bit geos library. Does that seem likely to you? | 11:29 |
cthibert | There are a couple places add the link for 64 bit in Ingres. I know I've only done the 32 bit in the current svn repo. | 11:30 |
FrankW | I did notice that -lgeos_c was missing in Jamdefs.i64_lnx - I'm not sure if there is other stuff missing. | 11:30 |
cthibert | Andrew says you made the change in Jamdefs, but there's one more file to change, sec I'll find it again. | 11:31 |
FrankW | Should things like unimapcompile really depend on geos? | 11:31 |
FrankW | Ah, cool | 11:31 |
cthibert | Hmm the other file is src/tools/port/shell_unix/mkdefault.sh but it doesn't seem to make a distinction between 32 and 64. | 11:33 |
FrankW | I'm a bit unclear what change I would make in mkdefault.sh. Just clone the int_lnx) line for i64_lnx? | 11:36 |
cthibert | No. It looks like mkdefault.sh is fine the way it is actually. It uses the same $suffix as int_lnx. There goes one potential fix... Hmm. Did you happen to install your geos lib in a 'custom' directory? | 11:39 |
FrankW | It is in a custom directory, yes. | 11:39 |
FrankW | But I *suspect* the issue is that unimapcompile is 32bit and I don't have a 32bit geos_c. | 11:39 |
FrankW | In Jamdefs.i64_lnx I put: | 11:40 |
FrankW | LDLIBMACH32 = -lpthread -lrt -lm -lc -lcrypt -ldl -lgcc_s -lgeos_c ; | 11:40 |
FrankW | LDLIBMACH = -lpthread -lrt -lm -lc -lcrypt -ldl -lgcc_s -lgeos_c ; | 11:40 |
FrankW | It frustrates me that the actual link lines used do not appear in the Jam logs. | 11:40 |
pboro | i64? isn't that itanium? | 11:42 |
pboro | you should be looking at a64_lnx if it's em64t or amd 64-bit | 11:42 |
FrankW | could be - I just guessed. | 11:42 |
pboro | I checked, i64_lnx is for itanium... I guess you should be using a64_lnx, what is your hw platform? | 11:43 |
FrankW | well, let me try again after updating Jamdefs.a64_lnx. | 11:43 |
cthibert | Your suspicion sounds plausible. Also, mkdefault.sh sets up the ldlibpath to /lib /usr/lib and /usr/local/lib so it may also be ignoring any custom library. Unfortunately, Jam is not my specialty and I'll be in tough to help with the jam debug files. | 11:43 |
FrankW | ok, rerunning build ... I'm not really sure how to establish what my hw platform is pboro. | 11:45 |
FrankW | Link /wrk/home/warmerda/ingres/server/tools/bin/lp32/salign | 11:46 |
FrankW | /usr/bin/ld: skipping incompatible /usr/local/lib/libgeos_c.so when searching for -lgeos_c | 11:46 |
FrankW | That would presumably be because it is 64bit. | 11:46 |
FrankW | So, to conclude, I will seemingly need a 32bit geos library. | 11:47 |
FrankW | I'll try and whip one up here. | 11:47 |
cthibert | Looks like it. | 11:47 |
pboro | FrankW, I don't know anyone who has Itanium... so I guess you're fine with a64_lnx :) | 11:47 |
FrankW | indeed | 11:47 |
cthibert | Can I help provide a 32bit geos? I've got ones compiled in FC11 and Ubuntu. | 11:49 |
FrankW | If you can make one available from Ubuntu I'll give it a try. | 11:50 |
cthibert | Okay, let me get it out (I'm in FC11 atm.) I'll figure out a way to make it available through http or sftp or if you have another preferred method? | 11:51 |
FrankW | fc11 might work fine. | 11:52 |
FrankW | If you can make it http fetchable for me that would be great. | 11:52 |
cthibert | okay, one sec. | 11:52 |
FrankW | I don't want to make you reboot - I can always build it myself. | 11:52 |
cthibert | Nope, I should be able to access the Ubuntu one without reboot, but if not I'll put the FC11 one up. If I can I'll put both up. | 11:53 |
cthibert | FrankW: Check your email for the link to the libraries. | 12:01 |
FrankW | cthibert: cool, thanks. | 12:02 |
cthibert | np | 12:04 |
*** atrofast has joined #ingres | 12:19 | |
FrankW | I seem to be making better progress on my build now, though I had to strip things down to .a files for the 32bit geos libraries to avoid runtime problems. | 12:29 |
FrankW | I don't really grasp how mixed 32/64bit environments normally work. | 12:29 |
atrofast | FrankW: usually you have /usr/local/lib/ for native libs (usually 64 bit) and /usr/local/lib32 for 32 bit libs | 12:35 |
atrofast | Then both paths are in LD_LIBRARY_PATH or ldconf | 12:36 |
atrofast | gcc/g++ or ld will search those paths, and find the compatible ones | 12:36 |
atrofast | It automatically skips incompatible architectures | 12:36 |
atrofast | At least taht's how it has worked in my experience :) | 12:36 |
FrankW | Perhaps the problem was that I didn't add /usr/local/lib32 to my LD_LIBRARY_PATH. | 12:41 |
atrofast | In some cases I've had problems where I had to add the dirs to BOTH ldconf AND LD_LIBRARY_PATH | 12:42 |
*** atrofast1 has joined #ingres | 12:52 | |
atrofast1 | Bah iPhone tethering is not terribly stable... Just got discoed | 12:53 |
*** atrofast has quit IRC | 12:59 | |
FrankW | cthibert: any thoughts on http://osgeo.pastebin.ca/1512711 ? | 13:01 |
FrankW | I believe this build has -L/usr/local/lib32 -lgeos_c -lgeos in LDLIBMACH32. | 13:01 |
* FrankW tries again being more explicit with link libraries. | 13:04 | |
cthibert | Hmm. This looks like it's geos that can't find itself. | 13:04 |
atrofast1 | FrankW: I just remembered, there is a change necessary for 64 bit machines | 13:04 |
FrankW | I suspect libgeos_c wasn't finding libgeos. | 13:04 |
FrankW | what change was that? | 13:04 |
atrofast1 | In one of the files that lists the libs needed, -lgeos needs to be added for the 64 bit build | 13:04 |
atrofast1 | I can't remember on top of my head which one... Let me see if I can find the patch I made | 13:05 |
FrankW | I have added them in Jamdefs.a64_lnx. | 13:05 |
cthibert | Looks like it. I wonder if it has anything to do with the softlinks in the tar file. I would have tarred them up as is, which would be pointing to /usr/local/lib. | 13:05 |
atrofast1 | Okay i think that ight be it | 13:05 |
FrankW | I stripped down my /usr/local/lib32 to just the two .a files. | 13:05 |
FrankW | I'm rerunning the build to see if it helps. | 13:05 |
FrankW | Now I get /usr/include/c++/4.3/iostream:77: undefined reference to `std::ios_base::Init::Init()' | 13:06 |
FrankW | 13:06 | |
FrankW | Likely a difference of versions between the FC system cthibert built on and mine. | 13:07 |
FrankW | Err, the ubuntu system I mean. | 13:07 |
FrankW | crap ... c++ libraries are a hassle in this regard. | 13:07 |
FrankW | I'll try building locally using -m32 and see if that works. | 13:08 |
FrankW | Loading iimerge ... | 13:46 |
FrankW | adupoint.c:(.text+0x47a): undefined reference to `initGEOS_r' | 13:46 |
FrankW | I'm not clear why this isn't picking up geos. | 13:46 |
*** atrofast1 has quit IRC | 13:46 | |
cthibert | Likely the error is coming from sooner than that. I've got logs that show the same thing. And the first failed link in still in unimapcompile. I'm working on a similar issue in FC11 atm with some updated build scripts from main. When I figure it out I'll see if it translates to your problem. | 14:04 |
pboro | FrankW, you are aware that running "jam -q" makes jam stop immediately when error occurs, aren't you? | 14:05 |
pboro | or do you use runbuild.sh? | 14:05 |
*** DerMeister has quit IRC | 14:16 | |
cthibert | I wasn't aware. Making the change to runbuild.sh to allow -q. Thanks pboro | 14:16 |
FrankW | I am using runbuild.sh and I did not know that. | 14:18 |
pboro | Alex gave that tip while I was debugging my Ingres build environment, made things a bit easier :) | 14:21 |
cthibert | Okay, quick commit to the spatial branch. If you svn up you can get the -q option FrankW | 14:25 |
* FrankW grabs and uses ... thanks! | 14:26 | |
FrankW | holy cow, it even gives a nice detailed error message with the full command text! | 14:27 |
pboro | yup :) | 14:27 |
cthibert | I found the file to change for the new scripts, but it doesn't exist in the current spatial branch. And the rest of the changes in what I'm working on aren't of the -q quick commit style. :) But hopefully the new -q option helps more. | 14:28 |
pboro | I hope it helps, I'm off for bed now. Nites guys! -> | 14:28 |
cthibert | Thanks pboro. night | 14:28 |
cthibert | I did notice a file /devsrc/ingres-geo/src/tools/port/shell_unix/shlibinfo.sh that also has linking options. Not entirely familiar with it, but it may be something to poke at. | 14:29 |
*** atrofast has joined #ingres | 14:32 | |
FrankW | I have altered mkdefault.sh after all and I think I'm making progress. | 14:32 |
cthibert | Okay, cool. I'm being paged for diner though so I'll have to leave. Let me know by email and I'll check in first thing tomorrow. | 14:34 |
FrankW | :-) | 14:35 |
*** cthibert has left #ingres | 14:35 | |
*** atrofast has quit IRC | 14:52 | |
*** mull has quit IRC | 16:09 | |
*** rossand has quit IRC | 16:39 | |
*** toumi01 has left #ingres | 17:18 | |
*** mull has joined #ingres | 17:48 | |
*** chenzhen has joined #ingres | 18:50 | |
*** chenzhen has left #ingres | 18:50 | |
*** mull has quit IRC | 18:53 | |
*** rossand has joined #ingres | 20:21 | |
*** ChanServ sets mode: +o rossand | 20:21 | |
*** toumi01 has joined #ingres | 21:01 | |
*** rossand has quit IRC | 21:06 | |
*** FrankW has quit IRC | 21:54 | |
*** Alex|off is now known as Alex| | 22:11 | |
*** Gerhard has joined #ingres | 22:34 | |
*** Alex| is now known as Alex|off | 22:56 | |
*** Alex|off is now known as Alex| | 23:08 | |
*** magnusg| has joined #ingres | 23:26 | |
*** Gerhard has quit IRC | 23:38 |
Generated by irclog2html.py 2.7 by Marius Gedminas - find it at mg.pov.lt!