Last Updated: 03/28/2011
NEXT EMAIL
From: Jeremy Nix Sent: Monday, March 28, 2011 7:16 AM To: Murphy, Shawn N. Cc: i2b2 users group; Kuttan, Rajesh
Subject: Re: Enhancement Request: Expose query status in CRC_QRY_getQueryMasterList_xxx
requests
Shawn/Raj,
Thanks for the explanation. That cleared up some confusion on the matter. I look forward to seeing this feature in action. On a basic level, it should provide an easy way of creating scheduled queries/reports out of i2b2.
Jeremy Nix Senior Application Developer Cincinnati Children's Hospital Medical Center
On 03/25/2011 03:30 AM, Murphy, Shawn N. wrote:
Hi Jeremy,
The multiple query instances are modeled to support reruns of the same query master. This allows a query to be set up (to look for some patients for a clinical trial perhaps) and then be rerun periodically. Every time the query is rerun, a new query instance will appear with its particular list of results. Like Raj says, the rerun feature has not been implemented yet, but when it is it will be way cool!
Thanks, Shawn.
From: Kuttan, Rajesh Sent: Thursday, March 24, 2011 4:07 PM To: Jeremy Nix Cc: i2b2 users group
Subject: Re: Enhancement Request: Expose query status in CRC_QRY_getQueryMasterList_xxx requests
Currently it is only supported from the server side(CRC_QRY_runQueryInstance_fromQueryMasterId), UI don't support a second instance of the query master.
Raj
On 3/24/11 3:33 PM, Jeremy Nix wrote:
Can you explain when you would have multiple query instance's for a given query master.
Jeremy Nix
Senior Application Developer Cincinnati Children's Hospital Medical Center
On 03/24/2011 03:01 PM, RAJESH KUTTAN wrote:
Re: Enhancement Request: Expose query status in CRC_QRY_getQueryMasterList_xxx requests
Hi
Jeremy, Its a 1 to many relation between qt_query_master and qt_query_instance and hence the query run status resides on the second level.
Thanks
Raj
On 3/24/11 1:33 PM, Jeremy Nix wrote:
Background:
The current architecture of CRC with respect to the data model can be simplified to the attached ER diagram. Information within these tables can be retrieved utilizing the following CRC cell operations:
CRC_QRY_getQueryMasterList_fromUserId ?> returns contents of qt_query_master table specific to given user CRC_QRY_getQueryMasterList_fromGroupId ?> returns contents of qt_query_master table specific to given group CRC_QRY_getQueryInstanceList_fromQueryMasterId ?> returns contents of qt_query_instance table specific to parent query master record CRC_QRY_getQueryResultInstanceList_fromQueryInstanceId ?> returns contents of qt_query_result_instance table specific to parent query instance record
With our own webclient, I like to show the status of the query to the user as a tooltip (along with other pieces of information like when the query was executed and completed). For me to do this, I have to first make a request for all query master's. Then, I have to iterate through all of those query master id's and make subsequent requests to get the query instance records for that query master. The reason I have to get the query instance information is that the query instance record contains the status of the query (along with the start/end date/times). This can present an overhead on our webclient. There are things I can do that could potentially improve this situation...like holding off that secondary request until the user actually hovers over a given query. This solution is simply shifting the burden from the initial page load time to the rendering time for the tooltip (before displaying the tooltip, a request to get the query instance information would have to be made).
I realize that this limitation is specific to our i2b2 webclient and how we are trying to make the information available to the end user. But, I do also see the potential for both the stock JavaScript based webclient and the Eclipse based client to display this same type of information to the user either in a tooltip or within some other UI control. Not to say that those clients aren't already displaying that information; they're just having to make the same additional webservice requests that I am.
??????????????
Idea #1:
??????????????
Add new element "query_status_type" to query_masterType complexType in CRC_PSM_OBJ.xsd. Look up the query status in the qt_query_instance table.
Pros:
- No additional columns required.
- Could be done quite easily from programming perspective with left outer join to qt_query_instance table. Cons:
- Like Idea #3, this idea will not work if my underlying assumption that the relationship between qt_query_master and qt_query_instance of 1 to 1 is incorrect.
??????????????
Idea #2:
??????????????
Include a status column in the qt_query_master table. Add new element "query_status_type" to query_masterType complexType in CRC_PSM_OBJ.xsd.
Pros:
- Least amount of changes required. Cons:
- Can't think of any
??????????????
Idea #3:
??????????????
Merge the qt_query_master and qt_query_instance tables together as one. This way the combined contains all pertinent information about the query request, query status, and start/end date/times.
Pros:
- Gets rid of unnecessary table. Cuts down on number insert/update statements executed with new query request. Cons:
- Much more progressive in the amount of changes required.
- I'm not 100% positive that there isn't a case where we would have multiple qt_query_instance records for a single qt_query_master record. If so, this solution would break that functionality.
These are just some ideas I had that could facilitate my request. I'm sure there may be others available. I just wanted to bring about the discussion to see if something like this could be addressed in, say, the 1.7 release. Also, if I am incorrect in the usage of qt_query_instance table, please let me know.
Thanks,
NEXT EMAIL
From: Mendis, Michael E. Sent: Saturday, March 26, 2011 2:11 PM To: Murphy, Shawn N.; James Scott; Mauro; Jack London Cc: James Law; Dan Connolly; i2b2 AUG Members; Peter Beninato
Subject: Re: web client patient set doesn't display
This has been resolved and a new webclient 1.5.2 has been posted on the i2b2.org/software page
mike
On 3/8/11 7:40 AM, Murphy, Shawn N. wrote:
Hi James,
We should be able to get a web client-side fix out for this pretty soon, a null is a valid way to account for the absence of an age.
As a temporary fix, I agree that negative numbers would make sense, although they may still appear in averages and reported in distributions.
Thanks, Shawn.
From: James Scott Sent: Tuesday, March 08, 2011 10:27 AM To: Mauro; Jack London Cc: James Law; Dan Connolly; i2b2 AUG Members members; Peter Beninato
Subject: RE: web client patient set doesn't display
Hi.
Since an age of zero (in years, integer) is a valid value for the newly?born, I'd suggest setting the NULL age to an agreed?upon nonsense value which could be easily excluded – perhaps a negative value. Then it would become a "training item of interest". This could be extended to other numeric values. The same idea would apply to mandatory character values where the source value is null – perhaps something like 'I2B2 UNKNOWN', for example.
Regards,
James Scott - OCTRI DW Lead Oregon Health & Science University
From: Mauro Sent: Tuesday, March 08, 2011 12:52 AM To: Jack London Cc: James Law; Dan Connolly; i2b2 AUG Members members; Peter Beninato
Subject: Re: web client patient set doesn't display
Thank you so much.
2011/3/7 Jack London
Problem solved!
Thank you Peter, Dan and Jim.
Jim, you pointed to the specific error that Dan alluded to in general: the age_in_years_num had unacceptable (null) values. This occurs where we had no birthdate (nulls ok for birth date). At least for the time being, I set the age_in_years_num to zero (it's an integer) and all is well.
The AUG listserv to the rescue, once again ...
Jack
On Mar 7, 2011, at 11:37 AM, James Law wrote:
We have also seen this behavior in the webclient that Dan describes, there may already be a jira to look at it.
Looks like some of the patients don't have age:
<param name="age_in_years_num" column="age_in_years_num"/>
James
Dan Connolly 3/7/2011 11:34 AM
On Mon, 2011-03-07 at 11:23 -0500, Jack London wrote:
we set jboss server log to DEBUG on a VM machine and recreated the problem. Below is a section of the log that shows that correct >> retrieval message (data for 5 patients) - but the web client just hangs and does not display the patient list data. Does anyone see any problem from the log?
We've seen these symptoms when we neglected to use the convention of using '@' rather than null > in data fields.
See Expanding Patient Recordset Tree View Hangs [12-02-2010] > in +https://community.i2b2.org/wiki/display/community/AUG+E-mail+ > +Repository > > p.s. I renew my request for a normal email archive so that I can point > directly to relevant threads and messages.
Dan
NEXT EMAIL
From: Russ Waitman Sent: Thursday, March 24, 2011 2:09 PM To: Keith Marsolo Cc: i2b2 AUG Members
Subject: Re: Anybody have experience to share regarding storage options for i2b2
Thanks Keith and James for the replies.
The flash is pricey (fusion-io) but if applied at the right point might dramatically improve response time and ETL loads in other situations. Actually, some of the slower flash boards (ocz) are comparable price to what I'm getting quoted for SAN storage but I am not sure the degree to which they are supported by HP. Apparently people also like flash cards because the power consumption is much lower than a bunch of spinning disks. We're going to spend some time seeing where our data winds up sitting and if there are smart places to really boost performance and manage growth. We'll share our plan as well.
Russ
Keith Marsolo 3/24/2011 10:55 AM
Russ,
We're going through a similar conversation here at CCHMC. We currently use a SAN solution. We're currently split between IBM and NetApp appliances. In the next few months we'll be going to a new NetApp appliance. Our infrastructure folks keep threatening to put us on a flash-based solution because we're too demanding, but I think we can do better by reorganizing our existing processes.
Plus, I don't want to pay for it.
We're in the middle of compiling and doing a sanity check on all our different ETL processes (not just i2b2). I can share some of that once it's complete, though it will probably be a few weeks. I'll try to send over the current infrastructure layout in the meantime.
Keith
On Mar 23, 2011, at 12:10 PM, Russ Waitman wrote:
Hi,
We're reevaluating storage options as we start to make our i2b2 installation widely available. We
have some overall space and growth things we need to address but are also concerned about
performance when it's successful and we have several concurrent users.
We're running things on SUSE Linux and the database is Oracle 10g.
Currently, the database servers are on HP DL180s with 2 six core 2.66 ghz xenon processors X5650s. Our storage is local to the server on 600 GB 15k drives though we have some slower SATA drives for backing up file dumps.
We're getting source data from Epic (~800 GB) and then loading it into an i2b2 schema on the same identified server. These drives are RAID10.
The data is then de-identified and put on a separate server from which it is accessed by i2b2 (on a separate virtualized server), drives are RAID5.
We're wondering what others are doing both to support the ETL load and also for the final warehouse for the customer.
One solution is to move to a SAN but the other options are to move to memory card based solutions like fusion-io +http://www.fusionio.com+
Are other people running on SANs and if so what kind of tuning has been done to improve performance? Any performance stories?
Anyone using memory/solid state solutions, specifically fusion IO? Their claim is their card shows up as a block device that bypasses all the storage I/O bottlenecks and runs between 100,000 to a million IOPS depending on the card and its size. From what we learned, you could also choose to just put key tables spaces for indices in the memory card storage. Cards run between 160 GB and 5.12 TB in size.
Russ Waitman Associate Professor Director of Medical Informatics Department of Biostatistics University of Kansas Medical Center
NEXT EMAIL
From: Mendis, Michael E. Sent: Thursday, March 24, 2011 6:22 AM To: Holden, Nick; i2b2 AUG Members
Subject: RE: Server vulnerability - any experiences?
Nick,
Thanks for the reply. Do please keep me informed if it does include a issue with the i2b2 software.
Thanks
mike
From: Holden, Nick Sent: Thu 3/24/2011 6:16 AM To: Mendis, Michael E.; i2b2 AUG Members
Subject: RE: Server vulnerability - any experiences?
Hi Mike,
It was a home-grown VM, so the root password was not the i2b2 standard one.
It's important to note that at this stage we don't know whether the attack vector was related to any element of i2b2 or its prerequisites. It may well have been a weakness in the core CentOS packages. It does seem that it wasn't as simple as someone guessing / cracking a user password over ssh however.
Nick
Nick Holden BRICCS Senior Database Officer Leicester Cardiovascular Biomedical Research Unit Glenfield Hospital
From: Mendis, Michael E. Sent: 24 March 2011 10:00 To: Holden, Nick; i2b2 AUG Members
Subject: RE: Server vulnerability - any experiences?
Nick,
I am sorry to hear that about the attack. Was the root password changed prior to the attack? If so, I would be very interested in understanding how the hacker was able to access. Otherwise, the root and all the accounts use a standard well published password. I will also add extra documentation in the install manual to recommend that users change root and system accounts.
Sorry
Mike
From: Holden, Nick Sent: Thu 3/24/2011 5:25 AM To: i2b2 AUG Members
Subject: Server vulnerability - any experiences?
Hi all,
We recently suffered an attack directed at a VM which had been set up as a short-life test i2b2 installation. Vanilla CentOS with i2b2 and prerequisites only. The hacker gained root access, overwrote sshd and became able to install additional programs and initiate port scanning attacks on other servers in our cluster. It seems unlikely that the attack was the result of a dictionary brute force attack. Investigations are continuing, but I wondered if anyone had encountered similar experiences recently and if they'd identified the vector of attack?
Thanks,
Nick
Nick Holden BRICCS Senior Database Officer Leicester Cardiovascular Biomedical Research Unit Glenfield Hospital
NEXT EMAIL
From: McMurry, Andrew J. Sent: Wednesday, March 23, 2011 7:25 PM To: i2b2 AUG Members Cc: Gene Hart; Dustin Key; Garth Arnold; David X Eastman
Subject: SAS scripts available for Ontology/ETL
Gene Hart et all presented a powerful opportunity yesterday for HMO users interested in i2b2/shrine.
Because the HMO VDW is standardized across many locations, the SAS scripts they have developed are highly reusable. Technically speaking, the opportunity to load up many HMOs may now be possible with relatively low effort.
Miraculously, KP North and Group Health are now in a pilot 2 node SHRINE HMO network with 15M patients.
The attached ZIP file includes SAS scripts and documentation for i2b2 HMO users. I HIGHLY encourage you to check it out.
Gene Hart, Dustin Key, Garth Arnold, David Eastman put these materials together and did the work. They are cc'd on this message and VERY generously shared these programs with the community.
Good day for i2b2, kudos gentleman!
andy
NEXT EMAIL
From: Kohane, Isaac Samuel Sent: Monday, March 21, 2011 9:35 PM To: Teresa Bosler Cc: jack.london; Murphy, Shawn N.; i2b2 AUG Members
Subject: Re: web client user documentation
Thanks Teresa, I just downloaded them.
On Mar 21, 2011, at 4:13 PM, Teresa Bosler wrote:
At UT Southwestern, I created some QuickTime Movies using stock out-of-the-box Apple QuickTime. Quicktime allows you to save automatically in multiple formats and produces all the http code along with it so all you do is drop it to the root of you web server and link to it. It is pretty much a no brainer solution to training users how to do something easy. Each movie took about 8 hours to script out and produce. I have placed them on my personal .me account at https://public.me.com/c.bosler. I will only leave them out there a couple of days so grab them while you can.
Teresa
Teresa Bosler, PMP EAC-Employee Advisor Council Member CTSA Liaison for Biomedical Informatics Department of Information Resources Academic Information Systems UT Southwestern Medical Center 5323 Harry Hines Blvd., BP4.106 Dallas, Texas 75390-9106
From: Donglu Xie Date: Fri, 18 Mar 2011 15:55:52 -0500 To: Teresa Bosler
Subject: FW: web client user documentation
FYI,
Do you want to contribute your voice record to them? I like it very much.
Donglu
From: Michael C Ford Sent: Friday, March 18, 2011 12:55 PM To: Jack London Cc: i2b2 AUG Members
Subject: Re: web client user documentation
Its on the website https://www.i2b2.org/software/index.html
03/18/2011 10:51 AM To i2b2 AUG Members
Subject web client user documentation
Does anyone have end user documentation for the i2b2 web client that they can share?
Jack London
UT Southwestern Medical Center The future of medicine, today.
NEXT EMAIL
From: Marc Sent: Friday, March 18, 2011 3:16 PM To: i2b2 AUG Members Cc: Audrey Hureau; GARCELON Nicolas
Subject: how to reuse DRG data in I2B2
Dear all, we are testing I2B2 here in Brittany ( France). We would like to load into the datawarehouse, data coming from our DRG system that contain diagnosis and procedures of inpatients. Do you have any experience about reusing DRG data in I2B2 infrastructure. We are wondering how to represent the patient pathway, i mean the different stays in the different clinical wards during hospitalization. Another question is related to the terminology versioning? For instance, how to keep data consistent from a previous version of ICD to a new one?
Thks a lot for your help
Marc
Here in frane Dr Marc CUGGIA (MD, PhD, MCU-PH) Inserm U936 Faculté de Médecine, Rue du Pr Léon Bernard 35043 Rennes cedex DIM - CHU Pontchaillou, rue H. Le Guilloux 35033 Rennes
NEXT EMAIL
From: Peter Beninato Sent: Thursday, March 17, 2011 7:30 PM To: i2b2 AUG Members
Subject: i2b2 wiki - modifier code - question(s)/suggestion(s)
Hi,
I read through some of the posting on the i2b2 wiki:
https://community.i2b2.org/wiki/display/DevForum/Modifiers+in+i2b2+Data+Model
https://community.i2b2.org/wiki/display/DevForum/Metadata+XML+for+Medication+Modifiers
One suggestion I have for improvement of the "Metadata XML for Medication Modifiers" document might be to add a representation of the records that are in the i2b2 table.
It would help my understanding, at least showing the c_basecode, and c_fullname/c_dimcode.
As I thought about what those rows would look like and I try to tie it into the "Modifiers in i2b2 Data Model" document I wanted to test my understanding...
The <TestID> in the Frequency section of the "Metadata XML for Medication Modifiers " document indicates "MED:FREQ".
<TestID>MED:FREQ</TestID>
This corresponds to the modifier_cd in the observation_Fact in the "Modifier i2b2 Data Model" document.
123|567|1|med:aspirin|20100404|MED:FREQ|T|QD|<null>
Then in the <EnumValues> there is a entry "Once per day" that maps to the "QD" in the tval_char column.
<Val description="Once per day">QD</Val> And voila! ???
Some other questions/suggestion:
1) Should the Val description tags in the ROUTE section of the "Metadata XML for Medication Modifiers" document also be between the ENUM tags?
2) There is a unique constraint on the concept_dimension (concept_path). It would seem that all the rows in the i2b2 metadata table would have the same c_basecode = "med:aspirin", but different c_metadataxml values for the different modifier codes. Is the idea that the different modifier codes would be under different branches in the ontology, so c_fullname/c_dimcode/concept_path are unique?
3) Could you expand the "Modifiers in i2b2 Data Model" document or create another document for examples of Pharmacy ID, PBM etc?
The info on the wiki is great and very helpful.
Thanks.
Peter Beninato - OCTRI DW Developer Oregon Health & Science University
NEXT EMAIL
From: Phillips, Lori C. Sent: Wednesday, March 16, 2011 11:07 AM To: Brian Ostasiewski; i2b2 AUG Members
Subject: RE: newline chars in observation_blob
Brian,
Not sure where you are seeing this, but in some plugins we had to temporarily replaceAll("/n", "\n")) prior to conversion to jaxb.
You may want to try the reverse ( replaceAll("\n", "/n") ) prior to your use of the blob.
Lori Phillips
From: Brian Ostasiewski Sent: Friday, March 11, 2011 12:32 PM To: i2b2 AUG Members
Subject: newline chars in observation_blob
It appears somewhere in the call chain newlines \n are being stripped out of the observation_blob text data, I'm guessing during xml return message generation. They're there in the database, but when the results come back to my plugin from the PDO query they're gone. Is this intentional? Configurable?
NEXT EMAIL
From: Kuttan, Rajesh Sent: Tuesday, March 15, 2011 4:05 PM To: Peter Beninato; i2b2 AUG Members
Subject: Re: i2b2 - 1.4 - query by value - question - between operator
Yes.
On 3/15/11 3:52 PM, Peter Beninato wrote:
Hi Raj,
Ok. To just check I have this correct, so since it is impossible to know that "< 2" might also be sub?zero, then it is not included.
Peter
From: RAJESH KUTTAN Sent: Tuesday, March 15, 2011 12:04 PM To: Peter Beninato; i2b2 AUG Members
Subject: Re: i2b2 - 1.4 - query by value - question - between operator
Hi Peter, Yes this is expected behavior, the query is taking into account the lower limit value. Raj
On 3/15/11 12:46 PM, Peter Beninato wrote:
Hi,
Have come to appreciate some of the functionality embedded in the "operator" encoding that populates tval_char.
For instance, if your lab result is beyond the limit of detection, and is represented in the source data as "< 2". I2b2 can be populated with Nval_num tval_char 2L
L = Less Than
Using the "Less than" operator in query?by?value, this captures both "equality" labs, and those below limit of detection (where clause snippet)
AND (( valtype_cd = 'N' AND nval_num < 2 AND tval_char IN
('LE','E'))OR ( valtype_cd = 'N' AND nval_num <= 2 AND tval_char ='L'))
My question has to do with the BETWEEN operator.
If I run a query where I'm looking for the value between 0 and 2, the where clause only looks for equality (tval_char = 'E')
AND nval_num BETWEEN 0 and 2AND tval_char = 'E')
Is this the expected behavior with no consideration for non?equality tval_char?
Peter Beninato - OCTRI DW Developer Oregon Health & Science University
NEXT EMAIL
From: Punna, Kiran Sent: Tuesday, March 15, 2011 2:41 PM To: i2b2 AUG Members
Subject: Reg: i2b2
Hi
Anybody having problems in uploading files to i2b2 workbench tool. Please help me with that.
Thank you,
Kiran Punna
Research Associate
NEXT EMAIL
From: Chan, Wayne Sent: Tuesday, March 15, 2011 1:36 PM To: 'Dat Q. Phan'; Phillip Reeder; i2b2 AUG Members
Subject: RE: 1.6 RC2
Thanks everyone. It turned out that the Apache Axis2 1.1 specified in the Hive Installation Guide was the problem. I got Johnny's axis.war and now I can get through to the i2b2 Administration – now I can continue with the setup (knock-on-woods).
Thanks again to everyone who helped, especially to Johnny & Phillip!
From: Dat Q. Phan Sent: Friday, March 11, 2011 4:39 PM To: Phillip Reeder; Chan, Wayne Cc: i2b2 AUG Members
Subject: RE: 1.6 RC2
First you want to prepare your i2b2 environment. I have attached some notes for installing this on Centos.
Then you want to follow the instructions provided in the documentation for i2b2 1.6 RC2.
Install required software (see attached).
install the data by following the "Data Installation Guide_1-6.pdf".
Install the PM cell by following the "PM_Installation_Guide_1-6.pdf".
Install the Ontology cell by following the "Ontology_Installation_Guide_1-6.pdf".
Install the Data Repository cell by following the "CRC_Installation_Guide_1-6.pdf".
Install the Workplace cell by following the "Workplace_Installation_Guide_1-6.pdf".
Again, let me know if you need any help.
Johnny Phan Programmer Analyst II UTHealth School of Biomedical Informatics
From: Phillip Reeder Sent: Friday, March 11, 2011 3:23 PM To: Chan, Wayne Cc: i2b2 AUG Members; Dat Q. Phan
Subject: Re: 1.6 RC2
Wayne, We have 1.6 RC2 running in a test environment at UT Houston. I believe we have some supplemental documentation/notes on the install process and I have asked Johnny to send it out.
Phillip
On Mar 11, 2011, at 1:11 PM, Chan, Wayne wrote:
Hi,
Has anybody successfully installed & running release 1.6 RC2. Thanks.
Wayne
NEXT EMAIL
From: Chan, Wayne Sent: Friday, March 11, 2011 4:42 PM To: 'Peter Beninato'; 'i2b2 AUG Members'
Subject: RE: getting stuck @ Section 4.1 "Administration of the i2b2 Project Management application" of PM_Installation_Guide_1-6.pdf
in the table pm_project_data – PROJECT_ID=Demo, PROJECT_NAME=i2b2 Demo, ..
in the table pm_project_user_roles – USER_ID=i2b2, PROJECT_ID=Demo, USER_ROLE_CD=ADMIN
From: Peter Beninato Sent: Friday, March 11, 2011 4:04 PM To: Chan, Wayne; i2b2 AUG Members Subject: RE: getting stuck @ Section 4.1 "Administration of the i2b2 Project Management application" of PM_Installation_Guide_1-6.pdf
Have you looked at the i2b2pm schema and database tables?
Do you have an entry in the table pm_project_data?
Do you see any entries in the table pm_project_user_roles for the user you hare trying to logon as?
From: Chan, Wayne Sent: Friday, March 11, 2011 11:20 AM To: 'i2b2 AUG Members'
Subject: RE: getting stuck @ Section 4.1 "Administration of the i2b2 Project Management application" of PM_Installation_Guide_1-6.pdf
Hi,
A colleague suggested that I turn off both the firewall & SELinux via
/usr/bin/system-config-securitylevel
/sbin/iptables --L
Then, the following error message by jboss is displayed following another failed attempt to login to admin:
14:02:09,794 ERROR [AxisRESTServlet] org.apache.axis2.AxisFault: edu.harvard.i2b2.pm.datavo.i2b2message.RequestMessageType$JaxbAccessorF_messageHeader cannot be cast to com.sun.xml.bind.v2.runtime.reflect.Accessor; nested exception is:
java.lang.ClassCastException: edu.harvard.i2b2.pm.datavo.i2b2message.RequestMessageType $JaxbAccessorF_messageHeader cannot be cast to com.sun.xml.bind.v2.runtime.reflect.Accessor
Is there a bug somewhere (remember, this is v.16RC2)?
Wayne
From: Chan, Wayne Sent: Thursday, March 10, 2011 4:17 PM To: 'i2b2 AUG Members'
Subject: getting stuck @ Section 4.1 "Administration of the i2b2 Project Management application" of PM_Installation_Guide_1-6.pdf Importance: High
Hi,
I have been stuck at Section 4.1 "Administration of the i2b2 Project Management application" of PM_Installation_Guide_1-6.pdf. I've installed all the data & PM, but when I got to the admin URL as following:
And tried to log in with username=i2b2 & pw=demouser, per the PM_Installation_Guide_1-6.pdf, all I got is the following complain:
I've been stuck at this for a few days, trying various things, and could use any helps any one can offer. Thanks in advance.
Wayne
NEXT EMAIL
From: Jeff Cowall Sent: Tuesday, March 08, 2011 3:41 PM To: Sebastian Mate Cc: i2b2 AUG Members
Subject: Re: Project and sub-project setup.
Thank you, Sebastian,
But this we can do already.
What we're working on is managing projects within projects, such that users in sibling projects can see their own data, and their parent's data, but not their sibling's.
The tables look designed to support this, and the design documents at least imply the functionality, but it's not brought clearly together anywhere.
Thanks again, though!
Best, Jeff
On 3/8/2011 at 1:53 PM, Sebastian Mate wrote:
Hello Jeff,
this might get you started: http://wwwcip.informatik.uni-erlangen.de/~sisemate/i2b2wizard/
A documentation is included in the ZIP file.
In the hope that it helps
Sebastian
Dipl.-Inf. Sebastian Mate Chair of Medical Informatics, Friedrich-Alexander-University Erlangen-Nuremberg, Germany Krankenhausstrasse 12, 91054 Erlangen
Am 08.03.2011 um 17:39 schrieb Jeff Cowall:
Is there documentation somewhere we've missed about how to setup multiple projects?
I know there multiple dimensions to the task, project definition (within a given hive),
project/subproject relationships, user/project roles, and more no doubt.
Reading the design documents reveals many clues, but since it spans multiple cells, it seems somewhat hard to not feel like there's something we're missing.
Thanks,
Jeff
Jeff Cowall Sr Data Architect University of Michigan Health System MCIT Clinical Research
NEXT EMAIL
From: Peter Beninato Sent: Tuesday, March 08, 2011 10:27 AM To: Kuttan, Rajesh; i2b2 AUG Members
Subject: RE: i2b2 - v1.4 - crc.properties - timeouts - JBOSS NOT{*}* shutting down
Hi Raj,
Thanks for the info.
Where is this "client's query tool option dialog" for setting the wait timeout?
Is this a JBOSS setting?
I'm not sure where this is? Please provide more info.
Thanks.
Peter
From: RAJESH KUTTAN Sent: Monday, March 07, 2011 7:41 PM To: Peter Beninato; i2b2 AUG Members
Subject: Re: i2b2 - v1.4 - crc.properties - timeouts - JBOSS NOT{*}* shutting down
Hi Peter, Please find the details below. Thanks Raj
On 3/7/11 6:37 PM, Peter Beninato wrote:
Hi,
In crc.properties file there is a line:
Setfinder JMS Queue transaction timeout properties edu.harvard.i2b2.crc.jms.small.timeoutsec=180
1) Does this correspond with the time that I see in the "Query Status" window?
No, this value just says how long the query can run in the small queue before it can be restarted from the medium queue.. If you want to increase the client wait time for big queries with 4/5 panels, then you can increase the wait timeout using the client's query tool option dialog .
2) Why when I change this to a more reasonable 1200, and redeploy via the ANT target, do I still timeout at 180?
Timeout at 180 is set from the client, you can increase it in the client side. If you change the property file, you can simply restart JBoss and no need to redeploy with ANT.
3) Why when I execute ./shutdown.sh --S, does JBOSS continue to run?
Sometime when the query is running, the shutdown.sh might not work. You can use kill command.
4) Is JBOSS continuing to run related to medium and large queues, and HOW DO I KILL THEM?
Something like this ps -ef | grep Main
Simply, just trying to boost the timeout, so that I can run a query with with 4/5 panels that takes longer than 180 seconds.
Peter Beninato - OCTRI DW Developer Oregon Health & Science University
NEXT EMAIL
From: Yves Thorrez Sent: Monday, March 07, 2011 10:20 AM To: i2b2 AUG Members
Subject: proxy configuration in Windows version of i2b2 client
Hi,
Is there any way to configure the proxy settings in the Windows version of the i2b2 client? I cannot add new features because we are behind a proxy.
When the client is built from source, it is possible to do this (via File|Preferences|General|Network Connections). However, while here I can retrieve the list of available features, there is a second problem: the error "Resulting configuration does not contain the platform" appears in the Updates dialog and I cannot proceed with Next. Searching the Web for this type of error has rendered nothing useful. Could perhaps something be wrong with my build process, such that the Eclipse platform is not included in my i2b2 runtime?
Thanks,
Yves
NEXT EMAIL
From: Murphy, Shawn N. Sent: Saturday, March 05, 2011 10:21 AM To: Holden, Nick; i2b2 AUG Members
Subject: RE: Integer values in i2b2 data other than lab users
Hi Nick,
I wrote up an application of the value schema to medications (as anther example) here: https://community.i2b2.org/wiki/display/DevForum/Metadata+XML+for+Medication+Mod ifiers
Thanks, Shawn.
From: Holden, Nick Sent: Thursday, March 03, 2011 5:24 AM To: i2b2 AUG Members
Subject: Integer values in i2b2 data other than lab users
Hi folks,
In trying to load our data into i2b2 we're confused by the scope for bringing value data, other than lab values.
For instance, blood pressure readings or similar measures. In the demo data ontology there are lab values available in such a way that a set value or range of values can be requested, whereas other data, such as age, is enumerated such that a selection of 'facts' represent age = 18, age = 19 and so on, and dragging various of these such facts into a query defines the required range.
Are these the only options?
We have data including blood pressure, weight and other measures which we want to be able to search on by specifying a maximum or minimum value, but they are not 'lab values' but neither is it straightforward to translate those integer values into a whole slew of Boolean variables holding facts.
How do we load such data into i2b2?
Thanks,
Nick
NEXT EMAIL
From: Peter Beninato Sent: Friday, March 04, 2011 6:31 PM To: Murphy, Shawn N.; i2b2 AUG Members
Subject: RE: i2b2 - web client - tool tip
Great! Another reason to upgrade.
From: Murphy, Shawn N. Sent: Friday, March 04, 2011 3:21 PM To: Peter Beninato; i2b2 AUG Members
Subject: RE: i2b2 - web client - tool tip
Hi Peter,
The tooltips are added in the 1.6 web client.
Thanks,
Shawn.
From: Peter Beninato Sent: Friday, March 04, 2011 5:29 PM To: i2b2 AUG Members
Subject: i2b2 - web client - tool tip
Hi,
Is there a way to get tool tips to work with the web client?
Using v1.4, and now have patient/observation counts working for ontology items. Rather than clutter up the display name with the count info, would like to have it available in the tool tip when hovering.
But it does not seem that tooltips work in web client.
Forgive my ignorance if this is classic shortcoming in web technology.
Thanks.
Peter Beninato - OCTRI DW Developer
Oregon Health & Science University
NEXT EMAIL
From: Arnold, Garth Sent: Friday, March 04, 2011 6:04 PM To: i2b2 AUG Members
Subject: RE: RAM requirement for the Application/Database server
Hello - I can't speak directly to needs for Oracle, but I find it useful to know the experience of others when sizing systems. In the hope that it's useful, here's my organization's i2b2 configuration and performance:
OS: Windows 2008 64-bit
db: SQL Server 2008 64-bit; disk use spread across 6 drives; drive space from iSCSI SAN
'hardware': virtual server, Intel Xeon-class CPUs; 4 vCPUs, 6GB RAM
memory usage: the i2b2 stack takes 300-400MB, SQL Server is configured to take 4GB; our DBAs suggest more RAM would be beneficial
- of facts: ~300 million (10 years of data on ~500,000 members)
- of users/usage: light enough that we don't worry much about concurrent queries
response time: generally in the 5-15 second range
Garth
Garth Arnold | MANAGER, TECHNICAL INFRASTRUCTURE
Group Health Research Institute
www.ghc.org
From: Murphy, Shawn N. Sent: Friday, March 04, 2011 5:04 AM To: Arvinder Choudhary; i2b2 AUG Members
Subject: RE: RAM requirement for the Application/Database server
We go with at least 32 GB on a 64 bit machine for our database servers.
Thanks, Shawn.
From: Arvinder Choudhary Sent: Thursday, March 03, 2011 10:36 AM To: i2b2 AUG Members; Murphy, Shawn N.
Subject: RAM requirement for the Application/Database server
Shawn and Team:
What will be the optimal size of RAM on the Application/Database server for 1-20 concurrent users. We use oracle 10g as the database with over half a billion observations.
Arvinder
Murphy, Shawn N. 3/1/2011 10:40 AM
It crashed when providing the question for "best in-it's-breed database". I think it was worried about the OFF switch.
From: Holden, Nick Sent: Tuesday, March 01, 2011 9:01 AM To: Murphy, Shawn N.; i2b2 AUG Members
Subject: RE: scaling issues with large concept hierarchy?
That's great Shawn, but I have to ask, how does the Partners i2b2 installation get on playing Jeopardy?
Nick
Nick Holden BRICCS Senior Database Officer Leicester Cardiovascular Biomedical Research Unit Glenfield Hospital
From: Murphy, Shawn N. Sent: 24 February 2011 17:15 To: Mendis, Michael E.; Dan Connolly; i2b2 AUG Members
Subject: RE: scaling issues with large concept hierarchy?
Hi Dan,
This is within the specification to have over 200,000 concepts in an i2b2 ontology.
There will be database-dependant and data-structure impacts on performance, many of which have been vetted in the past 3 weeks.
At Partners, with a fact table of 1.4 billion and a dimension table of about 150K concepts in a SQL Server databases, the simple queries take about 1-20 seconds.
Thanks, Shawn.
From: Mendis, Michael E. Sent: Thursday, February 24, 2011 9:42 AM To: Dan Connolly; i2b2 AUG Members
Subject: Re: scaling issues with large concept hierarchy?
It could be the indexes on the oracle 10g database, is this a enterprise or standard. I have heard of people having issues with the standard with huge amounts of data.
But to fix the jboss, chnge the memory size in JBOSS_HOME/bin/run.conf
mike
On 2/24/11 9:37 AM, Dan Connolly wrote:
We're exploring different organizations of concepts into hierarchies; some of them are sort of dumb structures built from some categorizations in Epic.
As a result, we have > 200,000 concepts. Is that within design expectations? Does anyone else have experience with that many concepts?
Yesterday, after bulk loading a bunch of data, bringing our observation_fact table to around 350 million rows, we saw performance issues in both navigating the concept tree and running queries by value (e.g. temperature above 38 degrees Celsius).
At least some of the problem was due to the JBoss middle tier; it was reporting out of memory problems:
2011-02-23 14:40:24,114 ERROR [org.jboss.ejb.txtimer.TimerImpl] Error invoking ejbTimeout: java.lang.RuntimeException: java.lang.OutOfMemoryError: PermGen space
And several jboss processes were running, oddly.
We suspect another source of poor performance is that Oracle 10g's Automatic Maintenance Tasks hadn't had a chance to analyze and rebuild indexes.
I'm new to Oracle tuning; I'm still learning what to expect in the way of performance. Any advice or experience reports are more than welcome.
NEXT EMAIL
From: Punna, Kiran Sent: Friday, March 04, 2011 3:48 PM To: i2b2 AUG Members
Subject: Reg:i2b2 importing files
Hi ,
Did anyone imported the CSV files to i2b2. Please help me with that. I am getting following error while uploading.
"Message error connecting project management cell "
Thank you,
Kiran
NEXT EMAIL
From: Jennifer Cai Sent: Friday, March 04, 2011 3:07 PM To: i2b2 AUG Members
Subject: webLogic and i2b2
Hi,
Does anyone have experience run i2b2 1.5 on WebLogic instead of JBOSS?
Thanks,
Jennifer
UT Southwestern Medical Center The future of medicine, today.
NEXT EMAIL
From: Jun Zhang Sent: Tuesday, March 01, 2011 5:15 PM To: i2b2 AUG Members
Subject: Re: i2b2 technical question
Dear AUG members,
We stalled the i2b2 core cells on a Linux server and got it to work with the (Harvard) demo data. We now want to see how to get the i2b2 to work with our own data. Is there a software for doing this?
We indexed our patient data with UMLS concepts and put the results (indexes) in a mySQL database. What do we need to do to:
1) Put the information that is in the mySQL database to the i2b2 required form (seems to be a Oracle database)?
2) Let i2b2 search program know about the resulting Oracle database so that it can search?
Thanks,
Jun Zhang University of Wisconsin-Milwaukee
NEXT EMAIL
From: Jeremy Nix Sent: Tuesday, March 01, 2011 3:43 PM To: Phillip Reeder Cc: Kuttan, Rajesh; i2b2 AUG Members
Subject: Re: hive cell/oracle password security approaches?
Phillip,
You can encrypt the Spring datasource passwords but it is a little more involved. Here's an article describing one possible way of doing this: http://mattluce.com/blog/2005/08/12/secure-passwordsfor-datasources-in-spring/ . The idea is to create a wrapper class around your datasource class that would decrypt the password at runtime.
Another method illustrated here (http://stackoverflow.com/questions/3423135/how-to-use-encryptedpassword-in-apache-basicdatasource) pushes the encrypted passwords to a properties file. At runtime, the property placeholders in the spring configuration file are replaced with the decrypted property values from the property file.
Something touched on in the 1st article that should probably be reiterated. As important as it may be to encrypt these plain text values, it is equally important to ensure that the file permissions are properly set.
Jeremy Nix Senior Application Developer Cincinnati Children's Hospital Medical Center
On 03/01/2011 01:39 PM, Phillip Reeder wrote:
We are using the method Raj sent to encrypt the database passwords in the XML files. What about the password that is in the CRCApplicationContext.xml?
<bean id="CRCDataSourceLookup" class="org.apache.commons.dbcp.BasicDataSource" destroy-method="close">
<property name="driverClassName" value="oracle.jdbc.driver.OracleDriver"/>
<property name="url" value="jdbc:oracle:thin:@localhost:1521:XE"/>
<property name="username" value="username"/>
<property name="password" value="password"/>
</bean>
Is there a way to encrypt this password as well? Or could it be moved to the crc-ds.xml?
Thanks,
Phillip
From: RAJESH KUTTAN Sent: Friday, September 10, 2010 5:21 PM To: Dan Connolly; i2b2 AUG Members
Subject: Re: hive cell/oracle password security approaches?
Attached the steps to encrypt the datasource password. Raj
On 9/10/10 2:16 PM, Dan Connolly wrote:
When we follow the i2b2 hive installation instructions, we end up with these datasource files that have passwords in them:
/usr/local/jboss-4.2.2.GA/server/default/deploy/crc-ds.xml /usr/local/jboss-4.2.2.GA/server/default/deploy/crc-jms-ds.xml /usr/local/jboss-4.2.2.GA/server/default/deploy/ont-ds.xml /usr/local/jboss-4.2.2.GA/server/default/deploy/pm-ds.xml /usr/local/jboss-4.2.2.GA/server/default/deploy/work-ds.xml
Does anyone have advice on how to secure those passwords?
The approach I'm using now is to create a dedicated 'jboss' user account, have jboss run as that user, and have those *-ds.xml files readable only by the jboss user.
I've been trying to get my head around the whole world of JBoss and J2EE... I gather these *-ds.xml files come from the Java Connection Architecture (JCA). Now I see something about JAAS and login-config.xml in JBoss documentation about JDBC connections http://docs.jboss.org/jbossas/docs/Server_Configuration_Guide/4/html/Connectors_on_JBossConfiguring_JDBC_DataSources.html. Does anybody have experience using that stuff?
NEXT EMAIL
From: Murphy, Shawn N. Sent: Tuesday, March 01, 2011 11:40 AM To: Holden, Nick; i2b2 AUG Members
Subject: RE: scaling issues with large concept hierarchy?
It crashed when providing the question for "best in-it's-breed database". I think it was worried about the OFF switch.
From: Holden, Nick Sent: Tuesday, March 01, 2011 9:01 AM To: Murphy, Shawn N.; i2b2 AUG Members
Subject: RE: scaling issues with large concept hierarchy?
That's great Shawn, but I have to ask, how does the Partners i2b2 installation get on playing Jeopardy?
Nick
Nick Holden BRICCS Senior Database Officer Leicester Cardiovascular Biomedical Research Unit Glenfield Hospital
From: Murphy, Shawn N. Sent: 24 February 2011 17:15 To: Mendis, Michael E.; Dan Connolly; i2b2 AUG Members
Subject: RE: scaling issues with large concept hierarchy?
Hi Dan,
This is within the specification to have over 200,000 concepts in an i2b2 ontology.
There will be database-dependant and data-structure impacts on performance, many of which have been vetted in the past 3 weeks.
At Partners, with a fact table of 1.4 billion and a dimension table of about 150K concepts in a SQL Server databases, the simple queries take about 1-20 seconds.
Thanks, Shawn.
From: Mendis, Michael E. Sent: Thursday, February 24, 2011 9:42 AM To: Dan Connolly; i2b2 AUG Members
Subject: Re: scaling issues with large concept hierarchy?
It could be the indexes on the oracle 10g database, is this a enterprise or standard. I have heard of people having issues with the standard with huge amounts of data.
But to fix the jboss, chnge the memory size in JBOSS_HOME/bin/run.conf
mike
On 2/24/11 9:37 AM, Dan Connolly wrote:
We're exploring different organizations of concepts into hierarchies; some of them are sort of dumb structures built from some categorizations in Epic. As a result, we have > 200,000 concepts. Is that within design expectations? Does anyone else have experience with that many concepts?
Yesterday, after bulk loading a bunch of data, bringing our observation_fact table to around 350 million rows, we saw performance issues in both navigating the concept tree and running queries by value (e.g. temperature above 38 degrees Celsius).
At least some of the problem was due to the JBoss middle tier; it was reporting out of memory problems:
2011-02-23 14:40:24,114 ERROR [org.jboss.ejb.txtimer.TimerImpl] Error invoking ejbTimeout: java.lang.RuntimeException: java.lang.OutOfMemoryError: PermGen space
And several jboss processes were running, oddly.
We suspect another source of poor performance is that Oracle 10g's Automatic
Maintenance Tasks hadn't had a chance to analyze and rebuild indexes.
I'm new to Oracle tuning; I'm still learning what to expect in the way of performance. Any advice or experience reports are more than welcome.
