Database Tasks error

[discofris]

I'm using the following on a winbox 2012R2:

• MC 3.4.2.8129
• JVM 1.8.0_111-b14
• PostgreSQL 9.6.1

When clicking the Database Tasks tab in the Settings I get an error:


Not really an issue for my setup I guess, but it got me wondering ... so I used the MC Client API to do a /databaseTasks GET Request and got the following Response back:

<com.mirth.connect.donkey.server.data.DonkeyDaoException>
    <detailMessage>
    org.postgresql.util.PSQLException: ERROR: column am.amcanorder does not exist
    Position: 427</detailMessage>
    <cause class="org.postgresql.util.PSQLException">
        <detailMessage>
      ERROR: column am.amcanorder does not exist
      Position: 427</detailMessage>
        <stackTrace>
            <trace>org.postgresql.core.v3.QueryExecutorImpl.receiveErrorResponse(QueryExecutorImpl.java:2270)</trace>
*** CUT ***
            <trace>java.lang.Thread.run(Unknown Source)</trace>
        </stackTrace>
        <suppressedExceptions class="java.util.Collections$UnmodifiableRandomAccessList" resolves-to="java.util.Collections$UnmodifiableList">
            <c class="list"/>
            <list/>
        </suppressedExceptions>
        <SQLState>42703</SQLState>
        <vendorCode>0</vendorCode>
        <__serverError>
            <m__mesgParts>
                <entry>
                    <char>P</char>
                    <string>427</string>
                </entry>
                <entry>
                    <char>R</char>
                    <string>errorMissingColumn</string>
                </entry>
                <entry>
                    <char>S</char>
                    <string>ERROR</string>
                </entry>
                <entry>
                    <char>C</char>
                    <string>42703</string>
                </entry>
                <entry>
                    <char>V</char>
                    <string>ERROR</string>
                </entry>
                <entry>
                    <char>F</char>
                    <string>parse_relation.c</string>
                </entry>
                <entry>
                    <char>L</char>
                    <string>3090</string>
                </entry>
                <entry>
                    <char>M</char>
                    <string>column am.amcanorder does not exist</string>
                </entry>
            </m__mesgParts>
            <verbosity>0</verbosity>
        </__serverError>
    </cause>
    <stackTrace>
        <trace>com.mirth.connect.server.util.DatabaseUtil.indexExists(DatabaseUtil.java:258)</trace>
*** CUT ***
        <trace>org.eclipse.jetty.util.thread.QueuedThreadPool$3.run(QueuedThreadPool.java:555)</trace>
        <trace>java.lang.Thread.run(Unknown Source)</trace>
    </stackTrace>
    <suppressedExceptions class="java.util.Collections$UnmodifiableRandomAccessList" resolves-to="java.util.Collections$UnmodifiableList">
        <c class="list"/>
        <list/>
    </suppressedExceptions>
</com.mirth.connect.donkey.server.data.DonkeyDaoException>

So I googled "column am.amcanorder does not exist" and found out it had something to do with the JDBC driver and PostgeSQL version 9.6.x:
http://comments.gmane.org/gmane.comp...sql.jdbc/22948

Currently not really a big issue I guess, but one never knows, right?

[discofris]

On that particular system I also had very often thoses error in my postgresql logs:

2016-11-08 11:12:02 CET [5624]: [467-1] user=,db=,app=,client= LOG:  could not fork autovacuum worker process: No error
2016-11-08 11:12:03 CET [5624]: [468-1] user=,db=,app=,client= LOG:  could not reserve shared memory region (addr=0000000001970000) for child 00000000000004BC: error code 487

When those occured, after x time the MC engine crashed. Solution was to restart engine and postgresql.

Tuning the autovacuum settings did not help.

Googling those errors pointed me into memory ASLR. Because disabling ASLR on Windows is also a security matter, I did not continue troubleshooting into that direction.

Reverted back to PostgreSQL 9.4.10, and had no more errors.

[narupley]

It appears this is a bug in the PostgreSQL JDBC driver:

• https://www.postgresql.org/message-i...mail.gmail.com
• https://github.com/pgjdbc/pgjdbc/pull/560
• https://github.com/pgjdbc/pgjdbc/pull/569

Try downloading the latest version of the driver here and replacing it in server-lib/database. I've created an issue for it: MIRTH-4078

[discofris]

Forgot to mention I already tried that with those drivers:
https://jdbc.postgresql.org/download....1212.jre7.jar (renamed those to the same name in the server-lib/database folder)

The error still persisted, so I realized it was not a good idea using PG version 9.6 and quickly installed 9.4

But still good to know it is certainly not advised to use PG 9.6 as a MC backend db on prod systems.

[narupley]

PostgreSQL 9.6 appears to work perfectly fine for me, so I still recommend it for production systems unless additional evidence comes up to the contrary.

After replacing the JAR make sure you restart the Mirth Connect server as well.

[discofris]

Did off course restart MC, on my 2 (win) test systems I concluded also it was production ready. But when installing it on customer site (only 4 channels!) ... the trouble began.

[narupley]

I haven't been able to reproduce those error logs you posted, so I don't know at this point whether they are isolated to your customer site or not. If you find any way to reproduce them on other systems, certainly let us know.