POST
|
Yes, I tried publishing directly from arcCatalog (right-click, share as, etc.) but when you publish that way, you don't have an option to include the data or not. It always includes the data in the .sd file when you do it that way. The locators themselves (the individual locators) are generated from layers in a file geodatabase in a subdirectory of the one where the locator was created (eg. /locator_dir/fgdb_dir). This should be irrelevant after the locator is created though, since all the data is indexed in the .lox file anyways. For the composite locators, the paths stored in the .loc file appear to be relative local paths to the individual locators (path = .). I am trying to publish directly from the server, so the paths are all valid. I did create a /published subfolder to put a copy of the locators in, and mapped this directory to my datastore in ArcGIS Server Manager (the mapping is publisher location d:\locators = server location d:\locators\published). I also created a datastore mapping for d:\locators as the publisher and server directory. For the composite locators, I now have a copy of the individual locators both in the root directory (D:\locators) as well as the published directory (D:\locators\published), so I'm not sure why it can't map the individual locators.
... View more
03-15-2016
01:06 PM
|
0
|
0
|
514
|
POST
|
In ArcGIS Server 10, we publish several individual locators (postal code, road, parcel) but also published a composite locator which included all the individually published locators. Since the composite locator was just a pointer to the other locators it published just fine. In 10.2.2 though the publishing process is very different. For all our services we are creating service definition files (.sd) to publish from. One of the options when creating the .sd file is to Include data in service definition when publishing. Since we are publishing the individual locators already, it is a waste of resources to publish the same locator data a second time so I have been trying to create the .sd file without including the data. When I go to publish the service though, it fails with an error 001487 - saying it can't update the server-side data location. I have created a data store that matches up the publisher directory with the server directory - this works fine for creating the individual locators without including their data in the .sd file. The problem seems to be that Server can't figure out that the data for the individual locators has already been uploaded and installed correctly. Is it not possible to create composite locators in the way we did in 10 - simply pointing the composite at existing individual locators? Is it just a matter of adding an additional data store configuration? I would rather not have to run multiple copies of the exact same locator, just to be able to include them in a composite locator. Any help would be greatly appreciated.
... View more
03-11-2016
01:48 PM
|
0
|
4
|
1978
|
POST
|
Brian Oevermann Thanks for your quick reply. I did a bit more digging through the xml file and think I've found the solution. I noticed that the MultiLineAddress section differed slightly from the NormalAddress - it was missing the OptionalFraction element (<elt ref="OptionalFraction" weight="40"/> ). I added that in and reloaded my locator and it seems to be working better now. Not sure why is worked okay with alpha suffixes like 'A' or 'B' but if it works, who am I to argue. Peter.
... View more
02-10-2016
08:53 AM
|
0
|
0
|
435
|
POST
|
Brian Oevermann, This thread is a bit old, but I was testing your locator style (we're using 10.2.2) and noticed that fractions in the house number don't seem to be recognized when using multi-line input. Multi-line works fine with other suffixes, like A,B or C - just not with fractions. I tried using 1/2 (i.e. 3 ascii characters) and ½ (ascii ½ ) - neither works. Single line input DOES recognize fractions and returns the correct results. Have you tested your style with fractions on multi-line inputs?
... View more
02-10-2016
08:19 AM
|
0
|
2
|
511
|
POST
|
I've been upgrading our existing 10.05 locators to 10.2.2 and thought everything was going well until I published our parcel service to Server 10.2.2. I'm experiencing unusual behaviour when submitting multiline locator requests both through the REST endpoint and SOAP (through ArcMap batch geocoding). What I'm seeing is that consistently, the first request for an address will return either the expected results or 'none'. The next request will return the opposite. This alternating behaviour is consistent and repeatable. I have tried rebuilding the locator using our previous 10.05 Single Address locator style as well as the one that comes with 10.2.2 and even tried the ones from 10.3.1 with the same results. This only seems to occur with our parcel data (it's the only one we publish using the single address style) - our roads locator based on dual range style works fine. It also works fine when using the single line input - just multiline isn't working. Has anyone else experienced this behaviour? Any workaround / fix?
... View more
02-08-2016
08:11 AM
|
0
|
1
|
1822
|
POST
|
We are implementing a new ArcGIS Server 10.2.2 infrastructure with 2 servers in one AGS site. Several map services will use cached map tiles. We would like each server to have a local copy of each of the caches (if we move the data to a shared location, we are forced to pay for backups which would be prohibitively expensive) but when I try to set up a cache directory through Manager, I get an error message from the other server: Failed to register the server directory 'local'. Server machine xxxxxx returned an error. 'Unable to access the server directory 'd:\xxxxxx\cache' The directory I'm trying to register has been created on both ArcGIS servers and I have assigned read/write permission to the directory for the ArcGIS Server account. Is it possible to configure ArcGIS server in this way, so that each server would look in it's own D drive for the cache?
... View more
12-10-2015
08:32 AM
|
0
|
2
|
3872
|
POST
|
Has there been any resolution to this problem? I am running into a similar (exact?) situation while trying to migrate some geoprocessing services based on custom DLL's from 10.05 to 10.2.2. I'm getting the invalid tool message for everything I publish. I even took the sample 'Calculate Area' VB code and compiled it as per the instructions (any CPU, registered it with 32bit and 64bit esriregasm.exe, etc.) and it works fine in ArcCatalog 10.2.2 and publishes fine to ArcGIS Server 10.2.2 but when the task is run, the 816 error is always returned. No other messages of any value are to be found anywhere. I was able to open the .rtl file (mentioned in previous post) and rerun it successfully. Is there a problem with trying to use custom DLL's on AGS 10.2.x 64bit? Any thoughts on where to look would be appreciated.
... View more
10-02-2015
11:12 AM
|
0
|
1
|
296
|
POST
|
Peter, Statement #3 is correct. All aliases will be treated as if they are exactly the same. Brad I meant to edit my post but you answered before I could 🙂 #3 should have read: Same as #1 AND Queens, Brooklyn and Manhatten can all be called New York #4 should be: Same as #2 AND order not at all important - New York or Queens or Brooklyn or Manhatten could have been the primary name. Every name is equal to every other name and has no bearing on how a city is found in the index. I'll say that #4 does NOT work as it represents our initial understanding of aliases which is disproven by my tests shown in the original post.
... View more
02-03-2014
11:13 AM
|
0
|
0
|
328
|
POST
|
It really doesn't matter the order of the aliases because they are all treated as equivalent but the first alias in the list is treated as the primary. Brad Sorry Brad, but I'm finding your answer a bit contradictory.... If the order doesn't matter because they are all equal, then there really isn't (shouldn't be) a need for a 'Primary' city name. A primary name would only be a necessity if other names had a relationship to it (ie. they are alternates for the first name, but not alternates for other names in the list). So... for clarity, which of these statements is the correct one: 1) when someone adds a new set of aliases to the template, they MUST put the primary name first followed by any aliases in random order? This will result in any of the 2-n aliases being equivalent to the first name but not other non-primary aliases. Ex. New York, Queens, Brooklyn, Manhatten are in a list of aliases; The primary name is New York but can also be called Queens, Brooklyn or Manhatten, but Queens is not an alias for Brooklyn or Manhatten (and vice versa). 2) when someone adds a new set of aliases to the template, the order has no relevance and all names in the list are all aliases for each other. Ex. same alias list as above. New York can be called Queens, Brooklyn or Manhatten AND Queens can also be called Brooklyn or Manhatten. 3) same statements as above except Queens could also be called New York (reversing the order - primary name is also an alias for alternate name) OR.... am I missing your intent completely. Just looking for clarity. Peter.
... View more
02-03-2014
10:51 AM
|
0
|
0
|
328
|
POST
|
it can be the first, second, ... or even last entry in the alias definition. If that's the case (order does not matter in the template) how does it know what order to put them in when you create a locator? Since, from my testing, the city stored with the spatial record has to come first, how does the locator know which city in an unordered list of aliases has to come first? Or, is it up to the developer to ensure that the first name in a group of aliases is the primary name and then all OTHER aliases can be in any order? Peter.
... View more
02-03-2014
08:27 AM
|
0
|
0
|
328
|
POST
|
It never hurts to dump the REST cache as well (http://servername/Arcgis/rest/admin ). Stopping and starting the service should invalidate the cache anyways, but we've had to manually clear the cache at times in order to see updates.
... View more
01-27-2014
09:19 AM
|
0
|
0
|
758
|
POST
|
Maybe this is old news to some people, but while trying to sort out some city alias name issues I ran across some unexpected behaviour in our single house locator (parcel data). From what I've read in the very limited documentation available, the alias lists are simple sets of equivalent names for an object (whether it's a street, state, city, etc.). I have not seen any documentation that states there should be a specific order for the aliases you create. However, from my testing, it appears that the FIRST alias in the set (for city aliases at least) must be the name that will be found in your dataset. Is this common knowledge and I've just missed seeing it in the documentation? I have tested this with several different cities / aliases and the behaviour is 100% repeatable. For example: the dataset contains an address 1417 Hazelwood Dr., Gorham and it geocodes correctly using this address. However, Gorham could also be called Thunder Bay, so we create an alias for it like so: <alias_def> <alt>THUNDER BAY</alt> <alt>GORHAM</alt> </alias_def> .... and we would expect the address 1417 Hazelwood Dr., Thunder Bay to now geocode correctly - but it still doesn't match. If we reverse the order of the two aliases, putting GORHAM first and THUNDER BAY second, then both variations of the address will geocode successfully. Just wondering if this is known/expected behaviour or have others run into the same issue? I didn't find anything related to the order of aliases when I searched the forum. Peter
... View more
01-22-2014
04:42 AM
|
0
|
7
|
925
|
POST
|
What I state is that the results shown using the ArcMap geocoding tool can be incorrect and unpredicatable. My screenshot shows that none of the addresses were matched automatically even though they have scores of 100%. The minimum score for automatic matching was 75%. Therefore the results contradict each other - if the minimum match score was achieved the geocoding was successful and this should be reflected in the totals of matched/tied/unmatched addresses but they are not. In addition, if you simply click on the Automatic Rematch button, you will often see the results change - sometimes from 100% unmatched to 100% matched. The geocoding service itself is returning the correct/expected results when the same addresses (that ArcMap failed to show as geocoded) are submitted through the REST interface, so we can be certain that it is not a problem with the ArcGIS Server service. ArcMap does not use REST - it uses SOAP and the traffic is in compressed binary format so we can't look at the traffic with tools like Fiddler2. I'm not saying the problem is with the REST service - in fact I doubt that it is - I think the problem lies in the logic of the geocoding GUI within arcMap. Our .NET and FLEX based applications which consume the REST services all appear to be working correctly - the JSON responses are formatted correctly and have all the information necessary to determine if a match was found or not. I would at least like to be able to pinpoint why this behaviour is occuring. At least then we can document it for our users, and hopefully provide a solution as well.
... View more
01-20-2014
06:31 PM
|
0
|
3
|
1667
|
POST
|
In the .loc.xml file for your locator, look for the <def name="NormalAddress"> section. The default order of the address components in this section is Number, Street Name <elt ref="House" weight="15"/> <elt ref="FullStreetName" weight="80" pre_separator="required" separator_list=".,;-"/> you should be able to switch these two lines around to match the structure of your Faroe Island addresses: <elt ref="FullStreetName" weight="80" pre_separator="required" separator_list=".,;-"/> <elt ref="House" weight="15"/> You might need to remove the pre_separator property since the street name is now the first component in the address
... View more
01-17-2014
09:45 AM
|
0
|
0
|
385
|
POST
|
I've been working on getting our parcel locator to return close matches on the same street (eg. input 100 main street, mytown which doesn't exist and have 101 main street, mytown returned instead of 100 main street, someothertown) and ran across some very strange behaviour. The behaviour is 100% repeatable and I have found the pattern. I'll try to explain it as best as I can. If you try to geocode an address of 21 King Street, Midland (which doesn't exist), you get the following returned: 251 King Street, Midland 261 King Street, Midland 291 King Street, Midland What the locator seems to do is this: -find all the addresses that start with the first number of the requested address - "2" in the example, AND end with the last number of the requested address - "1" in the example, BUT only if it has an extra number in between. An example with a larger number explains things further: Input 100 King Street and you get back: 1040 King Street 1050 King Street 1070 King Street The locator finds all the parcels that start with the first TWO numbers of the input address AND end with the last number of the input address BUT only if they are 4 digit numbers (one more than the input street number). Has anyone run into this behaviour before? Any ideas on how to overcome the problem? We have a dual range locator that works just fine (matches close house numbers) and it appears to have the same XML code for the house number logic.
... View more
01-17-2014
09:39 AM
|
0
|
1
|
1867
|
Title | Kudos | Posted |
---|---|---|
1 | 11-26-2013 06:49 AM | |
1 | 06-13-2019 12:17 PM |
Online Status |
Offline
|
Date Last Visited |
11-11-2020
02:23 AM
|