You are viewing a plain text version of this content. The canonical link for it is here.
Posted to dev@ranger.apache.org by Madhan Neethiraj <ma...@apache.org> on 2023/05/04 16:24:21 UTC
Re: Review Request 74398: RANGER-4190: Apache Ranger does not recover from a recoverable SQL exception
-----------------------------------------------------------
This is an automatically generated e-mail. To reply, visit:
https://reviews.apache.org/r/74398/#review225442
-----------------------------------------------------------
embeddedwebserver/scripts/ranger-admin-services.sh
Line 57 (original), 57 (patched)
<https://reviews.apache.org/r/74398/#comment314045>
java.security.egd seems unrelated to the DNS issue addressed in this patch. Is this necessary here?
Also, is this issue specific to OCI? Is there additional overhead in not caching positive name resoluions?
- Madhan Neethiraj
On April 15, 2023, 2:57 p.m., Ramachandran Krishnan wrote:
>
> -----------------------------------------------------------
> This is an automatically generated e-mail. To reply, visit:
> https://reviews.apache.org/r/74398/
> -----------------------------------------------------------
>
> (Updated April 15, 2023, 2:57 p.m.)
>
>
> Review request for ranger, Don Bosco Durai, Abhay Kulkarni, Madhan Neethiraj, Mehul Parikh, Nikhil P, Pradeep Agrawal, Ramesh Mani, Selvamohan Neethiraj, Sailaja Polavarapu, Subhrat Chaudhary, and Velmurugan Periasamy.
>
>
> Bugs: RANGER-4190
> https://issues.apache.org/jira/browse/RANGER-4190
>
>
> Repository: ranger
>
>
> Description
> -------
>
> It seems the Ranger instance is not able to connect Oracle ATP. This issue could be the loss of network connectivity to the load balancer which will connect to oracle ATP.
> java.sql.SQLRecoverableException: I/O Exception: Connection reset
>
> Internal Exception: java.sql.SQLRecoverableException: No more data to read from socket
> Error Code: 17410
> This issue could happen due to the loss of connectivity to the oracle ATP network load balancer.
> ADB network properties - https://docs.oracle.com/en/cloud/paas/autonomous-database/adbsa/connect-prepare-jdbc-thin.html#GUID-0A619DFE-5F5F-4302-BF43-B08AEDA05A6C https://docs.oracle.com/en/database/oracle/oracle-database/12.2/rilin/load-balancing-of-connections-to-oracle-rac-databases.html#GUID-06B4373D-1217-44E4-BD86-DE353DA657F6
> part of the Ranger start we need to add the below JVM parameters to disable the network cache by lowering its TTL to 0;-Djava.security.egd=file:///dev/urandom -Dnetworkaddress.cache.ttl=0 -Dnetworkaddress.cache.negative.ttl=0
>
>
>
> By default, the OpenJDK 8 InetAddress will cache failed DNS queries, referred to as "negative caching", for 10 seconds. It's recommended to disable both the positive and negative cache by setting them to 0 seconds, and rely on the DNS caching implemented inside OCI's DNS systems.
>
>
> java.security.Security.setProperty("networkaddress.cache.ttl", 0);
> java.security.Security.setProperty("networkaddress.cache.negative.ttl", 0);
> // (Optional) Force Java to use a specific DNS provider
> // By default it will use whatever lookup is defined in the system => files,nis,dns
> // as in /etc/nsswitch.conf on linux
> System.setProperty("sun.net.spi.nameservice.provider.1", "dns,sun");
> Or as an argument when invoking Java:
>
> Example:
> java -Dnetworkaddress.cache.ttl=0 -Dnetworkaddress.cache.negative.ttl=0 app.jar
>
>
> Diffs
> -----
>
> embeddedwebserver/scripts/ranger-admin-services.sh a76f0986c
>
>
> Diff: https://reviews.apache.org/r/74398/diff/1/
>
>
> Testing
> -------
>
>
> Thanks,
>
> Ramachandran Krishnan
>
>
Re: Review Request 74398: RANGER-4190: Apache Ranger does not recover from a recoverable SQL exception
Posted by Ramachandran Krishnan <ra...@gmail.com>.
> On May 4, 2023, 4:24 p.m., Madhan Neethiraj wrote:
> > embeddedwebserver/scripts/ranger-admin-services.sh
> > Line 57 (original), 57 (patched)
> > <https://reviews.apache.org/r/74398/diff/1/?file=2275867#file2275867line57>
> >
> > java.security.egd seems unrelated to the DNS issue addressed in this patch. Is this necessary here?
> >
> > Also, is this issue specific to OCI? Is there additional overhead in not caching positive name resoluions?
>
> Ramachandran Krishnan wrote:
> By default, the OpenJDK 8 InetAddress will cache failed DNS queries.We faced this issue due to the loss of connectivity to the oracle ATP network load balancer.Because of the cached DNS caching ,it was using the cached one and got the issue .I am not sure whether we are getting in another databases .
>
> Ramachandran Krishnan wrote:
> Let me revist once again for -Djava.security.egd=file:///dev/urandom
Madhan,
You are correct.
This (-Djava.security.egd=file:///dev/urandom) is no longer required if you are using JDK 8 or higher
For more info (search for random in the page):
https://docs.oracle.com/javase/8/docs/technotes/guides/security/enhancements-8.html
https://www.oracle.com/technetwork/java/javase/8-whats-new-2157071.html
- Ramachandran
-----------------------------------------------------------
This is an automatically generated e-mail. To reply, visit:
https://reviews.apache.org/r/74398/#review225442
-----------------------------------------------------------
On May 4, 2023, 5:38 p.m., Ramachandran Krishnan wrote:
>
> -----------------------------------------------------------
> This is an automatically generated e-mail. To reply, visit:
> https://reviews.apache.org/r/74398/
> -----------------------------------------------------------
>
> (Updated May 4, 2023, 5:38 p.m.)
>
>
> Review request for ranger, Don Bosco Durai, Abhay Kulkarni, Madhan Neethiraj, Mehul Parikh, Nikhil P, Pradeep Agrawal, Ramesh Mani, Selvamohan Neethiraj, Sailaja Polavarapu, Subhrat Chaudhary, and Velmurugan Periasamy.
>
>
> Bugs: RANGER-4190
> https://issues.apache.org/jira/browse/RANGER-4190
>
>
> Repository: ranger
>
>
> Description
> -------
>
> It seems the Ranger instance is not able to connect Oracle ATP. This issue could be the loss of network connectivity to the load balancer which will connect to oracle ATP.
> java.sql.SQLRecoverableException: I/O Exception: Connection reset
>
> Internal Exception: java.sql.SQLRecoverableException: No more data to read from socket
> Error Code: 17410
> This issue could happen due to the loss of connectivity to the oracle ATP network load balancer.
> ADB network properties - https://docs.oracle.com/en/cloud/paas/autonomous-database/adbsa/connect-prepare-jdbc-thin.html#GUID-0A619DFE-5F5F-4302-BF43-B08AEDA05A6C https://docs.oracle.com/en/database/oracle/oracle-database/12.2/rilin/load-balancing-of-connections-to-oracle-rac-databases.html#GUID-06B4373D-1217-44E4-BD86-DE353DA657F6
> part of the Ranger start we need to add the below JVM parameters to disable the network cache by lowering its TTL to 0;-Djava.security.egd=file:///dev/urandom -Dnetworkaddress.cache.ttl=0 -Dnetworkaddress.cache.negative.ttl=0
>
>
>
> By default, the OpenJDK 8 InetAddress will cache failed DNS queries, referred to as "negative caching", for 10 seconds. It's recommended to disable both the positive and negative cache by setting them to 0 seconds, and rely on the DNS caching implemented inside OCI's DNS systems.
>
>
> java.security.Security.setProperty("networkaddress.cache.ttl", 0);
> java.security.Security.setProperty("networkaddress.cache.negative.ttl", 0);
> // (Optional) Force Java to use a specific DNS provider
> // By default it will use whatever lookup is defined in the system => files,nis,dns
> // as in /etc/nsswitch.conf on linux
> System.setProperty("sun.net.spi.nameservice.provider.1", "dns,sun");
> Or as an argument when invoking Java:
>
> Example:
> java -Dnetworkaddress.cache.ttl=0 -Dnetworkaddress.cache.negative.ttl=0 app.jar
>
>
> Diffs
> -----
>
> embeddedwebserver/scripts/ranger-admin-services.sh a76f0986c
>
>
> Diff: https://reviews.apache.org/r/74398/diff/2/
>
>
> Testing
> -------
>
>
> Thanks,
>
> Ramachandran Krishnan
>
>
Re: Review Request 74398: RANGER-4190: Apache Ranger does not recover from a recoverable SQL exception
Posted by Ramachandran Krishnan <ra...@gmail.com>.
> On May 4, 2023, 4:24 p.m., Madhan Neethiraj wrote:
> > embeddedwebserver/scripts/ranger-admin-services.sh
> > Line 57 (original), 57 (patched)
> > <https://reviews.apache.org/r/74398/diff/1/?file=2275867#file2275867line57>
> >
> > java.security.egd seems unrelated to the DNS issue addressed in this patch. Is this necessary here?
> >
> > Also, is this issue specific to OCI? Is there additional overhead in not caching positive name resoluions?
>
> Ramachandran Krishnan wrote:
> By default, the OpenJDK 8 InetAddress will cache failed DNS queries.We faced this issue due to the loss of connectivity to the oracle ATP network load balancer.Because of the cached DNS caching ,it was using the cached one and got the issue .I am not sure whether we are getting in another databases .
Let me revist once again for -Djava.security.egd=file:///dev/urandom
- Ramachandran
-----------------------------------------------------------
This is an automatically generated e-mail. To reply, visit:
https://reviews.apache.org/r/74398/#review225442
-----------------------------------------------------------
On April 15, 2023, 2:57 p.m., Ramachandran Krishnan wrote:
>
> -----------------------------------------------------------
> This is an automatically generated e-mail. To reply, visit:
> https://reviews.apache.org/r/74398/
> -----------------------------------------------------------
>
> (Updated April 15, 2023, 2:57 p.m.)
>
>
> Review request for ranger, Don Bosco Durai, Abhay Kulkarni, Madhan Neethiraj, Mehul Parikh, Nikhil P, Pradeep Agrawal, Ramesh Mani, Selvamohan Neethiraj, Sailaja Polavarapu, Subhrat Chaudhary, and Velmurugan Periasamy.
>
>
> Bugs: RANGER-4190
> https://issues.apache.org/jira/browse/RANGER-4190
>
>
> Repository: ranger
>
>
> Description
> -------
>
> It seems the Ranger instance is not able to connect Oracle ATP. This issue could be the loss of network connectivity to the load balancer which will connect to oracle ATP.
> java.sql.SQLRecoverableException: I/O Exception: Connection reset
>
> Internal Exception: java.sql.SQLRecoverableException: No more data to read from socket
> Error Code: 17410
> This issue could happen due to the loss of connectivity to the oracle ATP network load balancer.
> ADB network properties - https://docs.oracle.com/en/cloud/paas/autonomous-database/adbsa/connect-prepare-jdbc-thin.html#GUID-0A619DFE-5F5F-4302-BF43-B08AEDA05A6C https://docs.oracle.com/en/database/oracle/oracle-database/12.2/rilin/load-balancing-of-connections-to-oracle-rac-databases.html#GUID-06B4373D-1217-44E4-BD86-DE353DA657F6
> part of the Ranger start we need to add the below JVM parameters to disable the network cache by lowering its TTL to 0;-Djava.security.egd=file:///dev/urandom -Dnetworkaddress.cache.ttl=0 -Dnetworkaddress.cache.negative.ttl=0
>
>
>
> By default, the OpenJDK 8 InetAddress will cache failed DNS queries, referred to as "negative caching", for 10 seconds. It's recommended to disable both the positive and negative cache by setting them to 0 seconds, and rely on the DNS caching implemented inside OCI's DNS systems.
>
>
> java.security.Security.setProperty("networkaddress.cache.ttl", 0);
> java.security.Security.setProperty("networkaddress.cache.negative.ttl", 0);
> // (Optional) Force Java to use a specific DNS provider
> // By default it will use whatever lookup is defined in the system => files,nis,dns
> // as in /etc/nsswitch.conf on linux
> System.setProperty("sun.net.spi.nameservice.provider.1", "dns,sun");
> Or as an argument when invoking Java:
>
> Example:
> java -Dnetworkaddress.cache.ttl=0 -Dnetworkaddress.cache.negative.ttl=0 app.jar
>
>
> Diffs
> -----
>
> embeddedwebserver/scripts/ranger-admin-services.sh a76f0986c
>
>
> Diff: https://reviews.apache.org/r/74398/diff/1/
>
>
> Testing
> -------
>
>
> Thanks,
>
> Ramachandran Krishnan
>
>
Re: Review Request 74398: RANGER-4190: Apache Ranger does not recover from a recoverable SQL exception
Posted by Ramachandran Krishnan <ra...@gmail.com>.
> On May 4, 2023, 4:24 p.m., Madhan Neethiraj wrote:
> > embeddedwebserver/scripts/ranger-admin-services.sh
> > Line 57 (original), 57 (patched)
> > <https://reviews.apache.org/r/74398/diff/1/?file=2275867#file2275867line57>
> >
> > java.security.egd seems unrelated to the DNS issue addressed in this patch. Is this necessary here?
> >
> > Also, is this issue specific to OCI? Is there additional overhead in not caching positive name resoluions?
By default, the OpenJDK 8 InetAddress will cache failed DNS queries.We faced this issue due to the loss of connectivity to the oracle ATP network load balancer.Because of the cached DNS caching ,it was using the cached one and got the issue .I am not sure whether we are getting in another databases .
- Ramachandran
-----------------------------------------------------------
This is an automatically generated e-mail. To reply, visit:
https://reviews.apache.org/r/74398/#review225442
-----------------------------------------------------------
On April 15, 2023, 2:57 p.m., Ramachandran Krishnan wrote:
>
> -----------------------------------------------------------
> This is an automatically generated e-mail. To reply, visit:
> https://reviews.apache.org/r/74398/
> -----------------------------------------------------------
>
> (Updated April 15, 2023, 2:57 p.m.)
>
>
> Review request for ranger, Don Bosco Durai, Abhay Kulkarni, Madhan Neethiraj, Mehul Parikh, Nikhil P, Pradeep Agrawal, Ramesh Mani, Selvamohan Neethiraj, Sailaja Polavarapu, Subhrat Chaudhary, and Velmurugan Periasamy.
>
>
> Bugs: RANGER-4190
> https://issues.apache.org/jira/browse/RANGER-4190
>
>
> Repository: ranger
>
>
> Description
> -------
>
> It seems the Ranger instance is not able to connect Oracle ATP. This issue could be the loss of network connectivity to the load balancer which will connect to oracle ATP.
> java.sql.SQLRecoverableException: I/O Exception: Connection reset
>
> Internal Exception: java.sql.SQLRecoverableException: No more data to read from socket
> Error Code: 17410
> This issue could happen due to the loss of connectivity to the oracle ATP network load balancer.
> ADB network properties - https://docs.oracle.com/en/cloud/paas/autonomous-database/adbsa/connect-prepare-jdbc-thin.html#GUID-0A619DFE-5F5F-4302-BF43-B08AEDA05A6C https://docs.oracle.com/en/database/oracle/oracle-database/12.2/rilin/load-balancing-of-connections-to-oracle-rac-databases.html#GUID-06B4373D-1217-44E4-BD86-DE353DA657F6
> part of the Ranger start we need to add the below JVM parameters to disable the network cache by lowering its TTL to 0;-Djava.security.egd=file:///dev/urandom -Dnetworkaddress.cache.ttl=0 -Dnetworkaddress.cache.negative.ttl=0
>
>
>
> By default, the OpenJDK 8 InetAddress will cache failed DNS queries, referred to as "negative caching", for 10 seconds. It's recommended to disable both the positive and negative cache by setting them to 0 seconds, and rely on the DNS caching implemented inside OCI's DNS systems.
>
>
> java.security.Security.setProperty("networkaddress.cache.ttl", 0);
> java.security.Security.setProperty("networkaddress.cache.negative.ttl", 0);
> // (Optional) Force Java to use a specific DNS provider
> // By default it will use whatever lookup is defined in the system => files,nis,dns
> // as in /etc/nsswitch.conf on linux
> System.setProperty("sun.net.spi.nameservice.provider.1", "dns,sun");
> Or as an argument when invoking Java:
>
> Example:
> java -Dnetworkaddress.cache.ttl=0 -Dnetworkaddress.cache.negative.ttl=0 app.jar
>
>
> Diffs
> -----
>
> embeddedwebserver/scripts/ranger-admin-services.sh a76f0986c
>
>
> Diff: https://reviews.apache.org/r/74398/diff/1/
>
>
> Testing
> -------
>
>
> Thanks,
>
> Ramachandran Krishnan
>
>