A change in Perl release 5.26.0 necessitates that the SIP_DIR variable
(default value "/opt/SIPServer") be exported to the environment in
oils_ctl.sh when attempting to start SIPServer. This change affects
Ubuntu 18.04 (Bionic Beaver) and Debian 10 (Buster) and any future
releases.
To verify the bug:
1. Install OpenSRF, Evergreen, and SIPServer on an affected O/S.
2. Start OpenSRF service.
3. Start SIPServer via oils_ctl.sh:
oils_ctl.sh -a start_sip
4. Notice the "Starting OILS SIP Server" message is printed with no
indication of failure.
5. Verify that the PID file exists: /openils/var/run/oils_sip.pid
6. Verify that no SIPServer processes are actually running:
pgrep -af SIPServer
The above should produce no output.
To test this patch:
1. Apply this patch and install oils_ctl.sh to /openils/bin/.
2. Assuming you have run the previous steps, delete the PID file:
rm /openils/var/run/oils_sip.pid
3. Start SIPServer via oils_ctl.sh:
oils_ctl.sh -a start_sip
4. Notice the "Starting OILS SIP Server" message is printed.
5. Verify that the PID file exists: /openils/var/run/oils_sip.pid
6. `pgrep -af SIPServer` should produce output.
It is safe to apply this patch on an unaffected O/S. Explicitly
adding SIP_DIR to PERL5LIB will have no detrimental effect on any O/S
currently supported by Evergreen.
Signed-off-by: Jason Stephenson <email address hidden>
Signed-off-by: Jason Boyer <email address hidden>
LP1937875: Remove GitHub reference from package.js
Removes the call to GitHub (required under legacy circumstances),
since it can cause some machines lacking SSH keys for GitHub to
fail when "npm install" is run for the AngularJS web client.
Signed-off-by: Adam Bowling <email address hidden>
Signed-off-by: Jason Boyer <email address hidden>
Add lp1923076-test-perl-scalar-return.t to the Perl live tests to test
a couple of back end calls affected by the changes in this branch.
These backend calls are those that are easily testable with Concerto
data.
Signed-off-by: Jason Stephenson <email address hidden>
Signed-off-by: Michele Morgan <email address hidden>
Signed-off-by: Mike Rylander <email address hidden>
LP#1923076: ensure length of empty array is treated as a number when retrieving hold counts
Thanks to Jason Stephenson for suggesting this solution.
Signed-off-by: Jeff Davis <email address hidden>
Signed-off-by: Jason Stephenson <email address hidden>
Signed-off-by: Michele Morgan <email address hidden>
Signed-off-by: Mike Rylander <email address hidden>