Trouble Report Ticket 34

Subject: Error Connecting to WSD Server
Creator: 28
Created: 2012-10-12 12:51:36
Transaction: 397
Parent: 0
MessageId: [email redacted]
Filename: 
ContentType: text/plain
ContentEncoding: none

Content: 
         >Category: mmtxT
         >Synopsis: Error Connecting to WSD Server
         >Confidential: no
         >Severity: serious
         >Priority: medium
         >Class: sw-bug
         >Submitter-Id: mmtxWeb
         >Release: 
         >Environment:
         >Operating System - MetaMap: 
         >OS Version - Ubuntu 12.04.1 LTS 32 bit
         >Java Version - Java 1.7
         >MySQL Version - 
         >Description:
         Frequent problems connecting to the WSD Server with the Java API.
         
         I start the POS tagger, WSD server, and Metamap server, all of which start successfully:
         
         >> ./bin/skrmedpostctl start
         $Starting skrmedpostctl: 
         started.
         >> ./bin/wsdserverctl start
         $Starting wsdserverctl: 
         started.
         loading properties file /../public_mm/WSD_Server/config/disambServer.cfg
         >> .bin/mmserver12
         /../public_mm/bin/SKRrun.12 /home/smokey/public_mm/bin/mmserver12.BINARY.Linux -L 2012 -Z 12
         Server options: [port(8066),accepted_hosts(['127.0.0.1'])]
         Berkeley DB databases (USAbase 2012AA strict model) are open.
         Static variants will come from table varsan in /../public_mm/DB/DB.USAbase.2012AA.strict.
         Derivational Variants: Adj/noun ONLY.
         Accessing lexicon /../public_mm/lexicon/data/lexiconStatic2012.
         Variant generation mode: static.
         port:8066
         
         I then run the testapi.sh script: 
         ./bin/testapi.sh mice
         
         This successfully runs and produces the desired output. However, when I request the WSD Server:
         ./bin/testapi.sh -y mice
         
         the program frequently hangs and the mmserver repeatedly produces the following output:
         
         
         ERROR in calling socket_client_open for WSD Server on host localhost and port 5554:
         error(system_error,system_error(SPIO_E_NET_CONNREFUSED))
         
         The log file (WSD_Server/log/WSD_Server.log) shows the following new entry (again, repeated multiple times):
         11 Oct 2012 11:54:31 ERROR [SocketResourcePool.java:129] wsd.util.SocketResourcePool - ind1.nlm.nih.gov
         
         
         After trying this several times, the connection might eventually work (and I have changed none of the configurations), and then proceed to process thousands of documents without difficulty. I will require greater consistency, so a solution to this problem is needed.
         
         Here are my specs: I am running the server and the testapi.sh scripts in two separate terminals on Ubuntu 12.04.1 LTS 32 bit in an Oracle VirtualBox VM. It has 3GB of memory.
         
         Thanks.
         >How-To-Repeat:
         
         >Unformatted:

Subject: 
Creator: 28
Created: 2012-10-12 16:54:30
Transaction: 402
Parent: 0
MessageId: [email redacted]
Filename: 
ContentType: text/html
ContentEncoding: none

Content: I was wondering how often you are restarting the WSD Server?   The current database used by default by the WSD Server requires 2gigabyte of memory to run.   Also, it can take up to minute for the server to initialize the databases before the server can be used.   The server is meant to be run once and then accessed multiple times by one or more instances of MetaMap.    Using 3gigabytes of memory may cause some slowness due to swapping.    

         

         It is possible to less memory by using the STI (Semantic Type Indexing) word sense disambiguation method instead of AEC (Automatically Extracted Corpus) but the results from STI are not as good.

         

         To use STI instead of AEC change the file public_mm/bin/SKRrun.12:

         

         export WSD_METHODS="['AEC_METHOD']"

         

         becomes:

         

         export WSD_METHODS="['STI_METHOD']"

         

         

         

         

         

         

Subject: [PublicMetamap #34] Error Connecting to WSD Server 
Creator: 1
Created: 2012-10-12 16:54:31
Transaction: 403
Parent: 0
MessageId: [email redacted]
Filename: 
ContentType: text/plain
ContentEncoding: none

Content: I was wondering how often you are restarting the WSD Server? The current
         database used by default by the WSD Server requires 2gigabyte of memory to run.
         Also, it can take up to minute for the server to initialize the databases
         before the server can be used. The server is meant to be run once and then
         accessed multiple times by one or more instances of MetaMap. Using 3gigabytes
         of memory may cause some slowness due to swapping.
         
         It is possible to less memory by using the STI (Semantic Type Indexing) word
         sense disambiguation method instead of AEC (Automatically Extracted Corpus) but
         the results from STI are not as good.
         
         To use STI instead of AEC change the file public_mm/bin/SKRrun.12:
         
         export WSD_METHODS="['AEC_METHOD']"
         
         becomes:
         
         export WSD_METHODS="['STI_METHOD']"
         

Subject: [PublicMetamap #34] Error Connecting to WSD Server 
Creator: 1
Created: 2012-10-12 16:54:32
Transaction: 404
Parent: 0
MessageId: [email redacted]
Filename: 
ContentType: text/plain
ContentEncoding: none

Content: I was wondering how often you are restarting the WSD Server? The current
         database used by default by the WSD Server requires 2gigabyte of memory to run.
         Also, it can take up to minute for the server to initialize the databases
         before the server can be used. The server is meant to be run once and then
         accessed multiple times by one or more instances of MetaMap. Using 3gigabytes
         of memory may cause some slowness due to swapping.
         
         It is possible to less memory by using the STI (Semantic Type Indexing) word
         sense disambiguation method instead of AEC (Automatically Extracted Corpus) but
         the results from STI are not as good.
         
         To use STI instead of AEC change the file public_mm/bin/SKRrun.12:
         
         export WSD_METHODS="['AEC_METHOD']"
         
         becomes:
         
         export WSD_METHODS="['STI_METHOD']"