-
Bug
-
Resolution: Not an Issue
-
P4
-
None
-
5.0u12
-
sparc
-
solaris_10
Hi,
Patch 118666-12 fell foul of our alternate root install testing on Solaris 10 sparc.
The command ran was:
luupgrade -t -n 5.10 -s /export/ 118666-12
Where 5.10 is the name of the alternate test partition.
The system under test has these process running on the live partition:
1842 /opt/SUNWcacao/lib/tools/launch -w /opt/SUNWcacao -r /var/opt/SUNWcacao/run/ret
1844 /usr/jdk/jdk1.5.0_07/bin/java -Xms4M -Xmx64M -classpath /opt/SUNWjdmk/5.1/lib/j
SUNWj5rt/install/postinstall calls (/a being the alternate partition)
2889: execve("/a/usr/jdk/instances/jdk1.5.0/bin/java", 0x00046DDC, 0x00046DFC) argc = 3
2889: argv: /a/usr/jdk/instances/jdk1.5.0/bin/java -client
2889: -Xshare:dump
That process subsequently sends a kill signal to the java process on the running partition:
2889: kill(1844, SIG#0) = 0
A requirement of live upgrade patching is that the patch being applied to the alternate partition cannot interfere with the live system. Either by changing files or sending signals to running processes.
This signal appears to have no ill effects, the target process still runs. Can someone please confirm what this signal does? If its effectivly a NOP then we can continue the release process for the patch.
Patch 118666-12 fell foul of our alternate root install testing on Solaris 10 sparc.
The command ran was:
luupgrade -t -n 5.10 -s /export/ 118666-12
Where 5.10 is the name of the alternate test partition.
The system under test has these process running on the live partition:
1842 /opt/SUNWcacao/lib/tools/launch -w /opt/SUNWcacao -r /var/opt/SUNWcacao/run/ret
1844 /usr/jdk/jdk1.5.0_07/bin/java -Xms4M -Xmx64M -classpath /opt/SUNWjdmk/5.1/lib/j
SUNWj5rt/install/postinstall calls (/a being the alternate partition)
2889: execve("/a/usr/jdk/instances/jdk1.5.0/bin/java", 0x00046DDC, 0x00046DFC) argc = 3
2889: argv: /a/usr/jdk/instances/jdk1.5.0/bin/java -client
2889: -Xshare:dump
That process subsequently sends a kill signal to the java process on the running partition:
2889: kill(1844, SIG#0) = 0
A requirement of live upgrade patching is that the patch being applied to the alternate partition cannot interfere with the live system. Either by changing files or sending signals to running processes.
This signal appears to have no ill effects, the target process still runs. Can someone please confirm what this signal does? If its effectivly a NOP then we can continue the release process for the patch.