Problems with latest splitter versions
Hi, I am getting the splitter from svn, compiling and running it unter Windows XP SP3 using JDK 1.6.0_25. There are 2 problems if I am using a splitter version newer than r161 (tryed until r167 so far). 1.) germany.osm.pbf from geofabrik cannot be split -------------------------------------------------- Invoking the splitter with java.exe -Xms256m -Xmx1536m -server -jar splitter.jar --resolution=14 --overlap=4000 --max-nodes=1200000 --max-areas=50 --geonames-file=cities1000.zip germany.osm.pbf The data is split into 76 areas, so there are 2 passes to go. First, everything looks fine, but then, at the and of the first pass, the splitter "falls asleep", the CPU load for the corresponding JVM drops to 0% and there ist no further progress (except the memory stats being printed periodically). The output looks like this (cut a lot of lines): cache= description= geonames-file=cities1000.zip legacy-mode=false mapid=63240001 max-areas=50 max-nodes=1200000 max-threads=1 (auto) mixed=false no-trim=false output-dir= overlap=4000 resolution=14 split-file= status-freq=120 write-kml= Elapsed time: 0s Memory: Current 247MB (1MB used, 246MB free) Max 1484MB Time started: Wed Mar 02 17:11:11 CET 2011 Map is being split for resolution 14: - area boundaries are aligned to 0x400 map units - areas are multiples of 0x800 map units wide and high Processing germany.osm.pbf Bounding box 5.864417 47.26543 15.050779999000001 55.14777 10.000.000 nodes processed... id=283576079 20.000.000 nodes processed... id=355168786 30.000.000 nodes processed... id=495365986 40.000.000 nodes processed... id=699656373 50.000.000 nodes processed... id=998291790 60.000.000 nodes processed... id=1127935431 1.000.000 ways processed... id=25049151 2.000.000 ways processed... id=28822644 Elapsed time: 1m 59s Memory: Current 247MB (20MB used, 227MB free) Max 1484MB 3.000.000 ways processed... id=34073254 4.000.000 ways processed... id=40593158 5.000.000 ways processed... id=49931588 6.000.000 ways processed... id=61188933 7.000.000 ways processed... id=83211109 8.000.000 ways processed... id=93106763 9.000.000 ways processed... id=101081600 100.000 ways processed... id=1124462 in 1 file Time: Wed Mar 02 17:14:00 CET 2011 Exact map coverage is (47.26541519165039,5.864403247833252) to (55.147762298583984,15.05077600479126) Trimmed and rounded map coverage is (47.28515625,5.8447265625) to (55.1513671875,15.029296875) Splitting nodes into areas containing a maximum of 1.200.000 nodes each... Area (47.5048828125,7.470703125) to (47.9443359375,8.7451171875) contains 663.137 nodes. DONE! [...] Area (53.8330078125,11.77734375) to (54.931640625,14.2822265625) contains 996.051 nodes. DONE! 76 areas: Area 63240001 covers (0x246800,0x4b000) to (0x24d800,0x4f800) DE-Dusseldorf [...] Area 63240076 covers (0x223000,0x63800) to (0x228800,0x70000) DE-Reutlingen Writing out split osm files Wed Mar 02 17:14:03 CET 2011 Processing 76 areas in 2 passes, 38 areas at a time Starting pass 1 of 2, processing 38 areas (63240001 to 63240038) Making SparseMultiMap Making SparseMultiMap Processing germany.osm.pbf Bounding box 5.864417 47.26543 15.050779999000001 55.14777 Making SparseMultiMap Making SparseMultiMap Making SparseMultiMap MAP occupancy: 832878 MAP occupancy: 158237 MAP occupancy: 8339 MAP occupancy: 546 MAP occupancy: 1652236 MAP occupancy: 326019 MAP occupancy: 19669 MAP occupancy: 2076 MAP occupancy: 2474023 MAP occupancy: 491531 MAP occupancy: 31585 MAP occupancy: 2861 MAP occupancy: 3272326 MAP occupancy: 668457 MAP occupancy: 53049 MAP occupancy: 6168 Elapsed time: 3m 59s Memory: Current 267MB (121MB used, 146MB free) Max 1484MB MAP occupancy: 4072409 MAP occupancy: 847909 MAP occupancy: 73212 MAP occupancy: 6470 MAP occupancy: 4870695 MAP occupancy: 1026785 MAP occupancy: 94770 MAP occupancy: 7750 MAP occupancy: 5673776 MAP occupancy: 1203597 MAP occupancy: 111531 MAP occupancy: 11096 10.000.000 nodes processed... id=283576079 [...] 60.000.000 nodes processed... id=1127935431 MAP occupancy: 37091591 MAP occupancy: 8145968 MAP occupancy: 704292 MAP occupancy: 58149 MAP occupancy: 37918126 MAP occupancy: 8301600 MAP occupancy: 722028 MAP occupancy: 58246 MAP occupancy: 38711472 MAP occupancy: 8485566 MAP occupancy: 744601 MAP occupancy: 58361 Writing ways Wed Mar 02 17:26:02 CET 2011 Making SparseMultiMap Making SparseMultiMap Making SparseMultiMap 1.000.000 ways processed... id=25049151 MAP occupancy: 798645 MAP occupancy: 184591 MAP occupancy: 15271 MAP occupancy: 1493 Elapsed time: 15m 59s Memory: Current 1484MB (705MB used, 779MB free) Max 1484MB 2.000.000 ways processed... id=28822644 [...] 7.000.000 ways processed... id=83211109 Elapsed time: 20m 2s Memory: Current 1484MB (928MB used, 556MB free) Max 1484MB MAP occupancy: 4795349 MAP occupancy: 1103602 MAP occupancy: 94312 MAP occupancy: 6736 MAP occupancy: 1 8.000.000 ways processed... id=93106763 9.000.000 ways processed... id=101081600 MAP occupancy: 5597978 MAP occupancy: 1283033 MAP occupancy: 110757 MAP occupancy: 8231 MAP occupancy: 1 Writing relations Wed Mar 02 17:32:01 CET 2011 100.000 ways processed... id=1124462 coords occupancy MAP occupancy: 39240504 MAP occupancy: 8605635 MAP occupancy: 752197 MAP occupancy: 58361 ways occupancy MAP occupancy: 5634593 MAP occupancy: 1292952 MAP occupancy: 111321 MAP occupancy: 8235 MAP occupancy: 1 Elapsed time: 22m 2s Memory: Current 1484MB (1144MB used, 340MB free) Max 1484MB Elapsed time: 24m 2s Memory: Current 1484MB (1144MB used, 340MB free) Max 1484MB Elapsed time: 26m 2s Memory: Current 1484MB (1144MB used, 340MB free) Max 1484MB Elapsed time: 28m 2s Memory: Current 1484MB (1144MB used, 340MB free) Max 1484MB Elapsed time: 30m 2s Memory: Current 1484MB (1144MB used, 340MB free) Max 1484MB At this point, nothing more happens, no second pass is done. 2.) File containing contour data cannot be processed ---------------------------------------------------- When generating my maps, I usually add data to draw contour lines. The file containing that data is available from http://osm.arndnet.de/contourdata.osm.gz (not my site!). It is in OSM format 0.5. When buildig a map, I am combining this data with the OSM data (please note that the contour data file contains node IDs starting at 2^31, so they collide with the IDs from "real" OSM data if used unchanged). But even when I am trying to split this file without additional OSM data, the same as above happens and the splitter stops at the end of the first and only pass: java.exe -Xms256m -Xmx1536m -server -jar splitter.jar --resolution=14 --overlap=4000 --max-nodes=1200000 --max-areas=50 --mixed --geonames-file=cities1000.zip contourdata.osm.gz cache= description= geonames-file=cities1000.zip legacy-mode=false mapid=63240001 max-areas=50 max-nodes=1200000 max-threads=1 (auto) mixed=true no-trim=false output-dir= overlap=4000 resolution=14 split-file= status-freq=120 write-kml= Elapsed time: 0s Memory: Current 247MB (1MB used, 246MB free) Max 1484MB Time started: Wed Mar 02 17:43:31 CET 2011 Map is being split for resolution 14: - area boundaries are aligned to 0x400 map units - areas are multiples of 0x800 map units wide and high Processing contourdata.osm.gz 2.500.000 nodes processed... 5.000.000 nodes processed... 7.500.000 nodes processed... 10.000.000 nodes processed... 12.500.000 nodes processed... 15.000.000 nodes processed... 17.500.000 nodes processed... in 1 file Time: Wed Mar 02 17:45:08 CET 2011 Exact map coverage is (47.2760796546936,5.877492427825928) to (55.02375841140747,14.999899864196777) Trimmed and rounded map coverage is (47.28515625,5.86669921875) to (55.0634765625,15.00732421875) Splitting nodes into areas containing a maximum of 1.200.000 nodes each... Area (47.5048828125,6.7236328125) to (49.1748046875,8.525390625) contains 958.841 nodes. DONE! [...] Area (52.822265625,10.17333984375) to (54.7119140625,14.47998046875) contains 885.962 nodes. DONE! 22 areas: Area 63240001 covers (0x23c000,0x54c00) to (0x247800,0x5ec00) DE-Koblenz [...] Area 63240022 covers (0x21c800,0x61000) to (0x22f800,0x6e800) DE-Stuttgart Writing out split osm files Wed Mar 02 17:45:10 CET 2011 Processing 22 areas in a single pass Starting pass 1 of 1, processing 22 areas (63240001 to 63240022) Making SparseMultiMap Making SparseMultiMap Processing contourdata.osm.gz Making SparseMultiMap Writing ways Wed Mar 02 17:45:14 CET 2011 Making SparseMultiMap Making SparseMultiMap Making SparseMultiMap Making SparseMultiMap Making SparseMultiMap Elapsed time: 1m 59s Memory: Current 1484MB (1090MB used, 394MB free) Max 1484MB MAP occupancy: 783578 MAP occupancy: 190519 MAP occupancy: 25903 MAP occupancy: 1545344 MAP occupancy: 394559 MAP occupancy: 60097 MAP occupancy: 2398068 MAP occupancy: 533001 MAP occupancy: 68931 2.500.000 nodes processed... MAP occupancy: 3315227 MAP occupancy: 615842 MAP occupancy: 68931 MAP occupancy: 4195362 MAP occupancy: 728752 MAP occupancy: 75886 5.000.000 nodes processed... [...] [...] Elapsed time: 7m 59s Memory: Current 1484MB (1197MB used, 287MB free) Max 1484MB MAP occupancy: 16438198 MAP occupancy: 3986436 MAP occupancy: 575366 MAP occupancy: 17300217 MAP occupancy: 4121087 MAP occupancy: 578696 17.500.000 nodes processed... coords occupancy MAP occupancy: 17896316 MAP occupancy: 4149555 MAP occupancy: 578696 ways occupancy MAP occupancy: 180407 MAP occupancy: 43178 MAP occupancy: 6700 MAP occupancy: 21 MAP occupancy: 1 Elapsed time: 9m 59s Memory: Current 1484MB (1013MB used, 471MB free) Max 1484MB Elapsed time: 11m 59s Memory: Current 1484MB (1014MB used, 470MB free) Max 1484MB Elapsed time: 13m 59s Memory: Current 1484MB (1014MB used, 470MB free) Max 1484MB Elapsed time: 15m 59s Memory: Current 1484MB (1014MB used, 470MB free) Max 1484MB ***** Full GC ***** Elapsed time: 18m 1s Memory: Current 1484MB (923MB used, 561MB free) Max 1484MB Elapsed time: 20m 1s Memory: Current 1484MB (924MB used, 560MB free) Max 1484MB Elapsed time: 22m 1s Memory: Current 1484MB (924MB used, 560MB free) Max 1484MB and so on. I've then converted the data to OSM format 0.6 and sorted it (using --migrate and --sort-0.6 for osmosis) and stored it as .osm.bz2 and .osm.pbf. When I am trying to split these files, I am getting an exception: java.exe -Xms256m -Xmx1536m -server -jar splitter.jar --resolution=14 --overlap=4000 --max-nodes=1200000 --max-areas=50 --geonames-file=cities1000.zip contourdata06.osm.bz2 cache= description= geonames-file=cities1000.zip legacy-mode=false mapid=63240001 max-areas=50 max-nodes=1200000 max-threads=1 (auto) mixed=false no-trim=false output-dir= overlap=4000 resolution=14 split-file= status-freq=120 write-kml= Elapsed time: 0s Memory: Current 247MB (1MB used, 246MB free) Max 1484MB Time started: Wed Mar 02 19:28:48 CET 2011 Map is being split for resolution 14: - area boundaries are aligned to 0x400 map units - areas are multiples of 0x800 map units wide and high Processing contourdata06.osm.bz2 2.500.000 nodes processed... 5.000.000 nodes processed... 7.500.000 nodes processed... Elapsed time: 2m 0s Memory: Current 247MB (81MB used, 166MB free) Max 1484MB 10.000.000 nodes processed... 12.500.000 nodes processed... 15.000.000 nodes processed... Elapsed time: 4m 0s Memory: Current 247MB (61MB used, 186MB free) Max 1484MB 17.500.000 nodes processed... in 1 file Time: Wed Mar 02 19:34:21 CET 2011 Exact map coverage is (47.2760796546936,5.877492427825928) to (55.02375841140747,14.999899864196777) Trimmed and rounded map coverage is (47.28515625,5.86669921875) to (55.0634765625,15.00732421875) Splitting nodes into areas containing a maximum of 1.200.000 nodes each... Area (47.5048828125,6.7236328125) to (49.1748046875,8.525390625) contains 958.841 nodes. DONE! [...] Area (52.822265625,10.17333984375) to (54.7119140625,14.47998046875) contains 885.962 nodes. DONE! 22 areas: Area 63240001 covers (0x23c000,0x54c00) to (0x247800,0x5ec00) DE-Koblenz [...] Area 63240022 covers (0x21c800,0x61000) to (0x22f800,0x6e800) DE-Stuttgart Writing out split osm files Wed Mar 02 19:34:24 CET 2011 Processing 22 areas in a single pass Starting pass 1 of 1, processing 22 areas (63240001 to 63240022) Making SparseMultiMap Making SparseMultiMap Processing contourdata06.osm.bz2 Exception in thread "main" java.lang.ArrayIndexOutOfBoundsException: Start index (-30655339) is negative at it.unimi.dsi.fastutil.Arrays.ensureFromTo(Arrays.java:45) at it.unimi.dsi.fastutil.objects.ObjectArrays.ensureFromTo(ObjectArrays.java:351) at it.unimi.dsi.fastutil.objects.ObjectArrays.fill(ObjectArrays.java:313) at it.unimi.dsi.fastutil.objects.ObjectArrayList.size(ObjectArrayList.java:308) at uk.me.parabola.splitter.SparseInt2ShortMapInline.resizeTo(SparseInt2ShortMapInline.java:96) at uk.me.parabola.splitter.SparseInt2ShortMapInline.put(SparseInt2ShortMapInline.java:123) at uk.me.parabola.splitter.SparseInt2ShortMultiMap$Inner.put(SparseInt2ShortMultiMap.java:81) at uk.me.parabola.splitter.SparseInt2ShortMultiMap.put(SparseInt2ShortMultiMap.java:31) at uk.me.parabola.splitter.SplitProcessor.writeNode(SplitProcessor.java:208) at uk.me.parabola.splitter.SplitProcessor.processNode(SplitProcessor.java:118) at uk.me.parabola.splitter.OSMParser.endElement(OSMParser.java:243) at uk.me.parabola.splitter.AbstractXppParser.parse(AbstractXppParser.java:57) at uk.me.parabola.splitter.Main.processMap(Main.java:399) at uk.me.parabola.splitter.Main.writeAreas(Main.java:355) at uk.me.parabola.splitter.Main.split(Main.java:188) at uk.me.parabola.splitter.Main.start(Main.java:116) at uk.me.parabola.splitter.Main.main(Main.java:105) Both files (the original file in OSM format 0.5 and the converted .osm.bz2 in OSM format 0.6) could be split without a problem using splitter r161. Any ideas what is going wrong there? Anything I could try? Michael
Le 03/03/2011 00:36, Michael Prinzing a écrit : Hi,
1.) germany.osm.pbf from geofabrik cannot be split --------------------------------------------------
MAP occupancy: 1 Elapsed time: 22m 2s Memory: Current 1484MB (1144MB used, 340MB free) Max 1484MB Elapsed time: 24m 2s Memory: Current 1484MB (1144MB used, 340MB free) Max 1484MB Elapsed time: 26m 2s Memory: Current 1484MB (1144MB used, 340MB free) Max 1484MB Elapsed time: 28m 2s Memory: Current 1484MB (1144MB used, 340MB free) Max 1484MB Elapsed time: 30m 2s Memory: Current 1484MB (1144MB used, 340MB free) Max 1484MB
At this point, nothing more happens, no second pass is done.
Any ideas what is going wrong there? Anything I could try?
I reported two days ago exactly the same problem to Steve. I sent him yesterday some debug info. Francois --
On Wed, Mar 2, 2011 at 5:36 PM, Michael Prinzing <mipri@gmx.net> wrote:
Hi,
I am getting the splitter from svn, compiling and running it unter Windows XP SP3 using JDK 1.6.0_25. There are 2 problems if I am using a splitter version newer than r161 (tryed until r167 so far).
1.) germany.osm.pbf from geofabrik cannot be split --------------------------------------------------
Invoking the splitter with
java.exe -Xms256m -Xmx1536m -server -jar splitter.jar --resolution=14 --overlap=4000 --max-nodes=1200000 --max-areas=50 --geonames-file=cities1000.zip germany.osm.pbf
The data is split into 76 areas, so there are 2 passes to go.
BTW, there's no need to do two passes. Unlike prior versions, this version of the splitter will handle as many areas as you want to throw at it. I've done 6000 at once on my 4GB ram desktop. Frantisek Mantlik tracked down and fixed the problem with threads=1.
2.) File containing contour data cannot be processed ----------------------------------------------------
When generating my maps, I usually add data to draw contour lines. The file containing that data is available from http://osm.arndnet.de/contourdata.osm.gz (not my site!). It is in OSM format 0.5.
When buildig a map, I am combining this data with the OSM data (please note that the contour data file contains node IDs starting at 2^31, so they collide with the IDs from "real" OSM data if used unchanged). But even when I am trying to split this file without additional OSM data, the same as above happens and the splitter stops at the end of the first and only pass:
This won't work. The splitter is written in Java, where most standard Java classes and collections don't support indices that are 64 bits. This means that ID's more than 2**31 are too big and will be treated as negative by Java and blow up this code. The current array resize logic further limits this to node ID's less than somewhere around 1,750,000,000, but I just committed a fix to trunk that I think will extend this range to 1--1,999,999,999. You could help me find and fix any other issues if you could supply a small file with node ID's around 1,950,000,000.
Exception in thread "main" java.lang.ArrayIndexOutOfBoundsException: Start index (-30655339) is negative at it.unimi.dsi.fastutil.Arrays.ensureFromTo(Arrays.java:45) at it.unimi.dsi.fastutil.objects.ObjectArrays.ensureFromTo(ObjectArrays.java:351) at it.unimi.dsi.fastutil.objects.ObjectArrays.fill(ObjectArrays.java:313) at it.unimi.dsi.fastutil.objects.ObjectArrayList.size(ObjectArrayList.java:308)
Scott
Hi Micheal, I tried your data to verify that it can be processed with my patched version in the memory_optimization branch. Unfortunately it seems that splitter has another problem not related to memory or high node id values: During the 1st phase, the program stops reading nodes too early. Result: The algorithm in splittableArea.split doesn't work because it uses a far too small value for the nodes: it counts only the 1st 145 nodes and stops counting when the 1st way is processed. I tried also a smaller sample from OSM with the same result: http://api.openstreetmap.org/api/0.6/map?bbox=11.54,48.14,11.543,48.145 Using parameter --legacy-mode=true changed it a little bit, but I think I think this is an error, so I'll try to find a fix for it. Am I right? Ciao, Gerd -- View this message in context: http://gis.638310.n2.nabble.com/Problems-with-latest-splitter-versions-tp608... Sent from the Mkgmap Development mailing list archive at Nabble.com.
participants (4)
-
frmas -
GerdP -
Michael Prinzing -
Scott Crosby