Skip to content

Production Installation Instructions

Shane Walker edited this page Nov 8, 2024 · 2 revisions

These instructions are written for Gemini South and semester 2024A, so adjust accordingly.

  • As I go, I remove older installations in /home/software and the old database files in /mount/petrohue/odbhome/ugemini/spdb/spdb.active/. For example, we no longer need to keep around 2023A.1.1 or 2023A.1.2.

  • Turn off the cron jobs, editing the crontab with crontab -e and commenting out at least the entries that are likely to execute during the installation.

  • Turn off the trpc bundle via the telnet admin interface (telnet gsodb 8224): locate the trpc bundle from the listing you get with lb (#70 in this case) and stop 70. This prevents the OT from connecting and updating the database.

  • Turn off the edu.gemini.wdba.xmlrpc.server bundle (#67) in the hopes that this will prevent the seqexec from connecting. I'm not sure if this is enough actually or if it will do what I hope it will. At any rate, it won't hurt anything to turn it off.

  • At this point normally I'd create a backup directory manually in /mount/petrohue/odbhome/ugemini/spdb/spdb.archive/20231212 and then via the telnet prompt exportXML /mount/petrohue/odbhome/ugemini/spdb/spdb.archive/20231212. Because I'm doing this via Keeper though, I'll execute the cron task instead and let that finish:

/usr/bin/curl -sS http://localhost:8442/cron/archive >> /home/software/install/log/cron.out 2>&1
  • Now we stop the ODB using the telnet admin with stop 0. Sometimes the database fails to shutdown cleanly and we have to kill it from the command line though.

  • There's some manual maintenance to move the VCS and program key information over from the previous version:

cd /mount/petrohue/odbhome/ugemini/spdb/spdb.active
mkdir 2024A.1.1
cd 2023B.1.1
cp -r keychain.ser keyserver vcs vcs.h2.db vcs.trace.db ../2024.1.1
  • Back to the new installation: cd ~/spdb_2024A.1.1.1_linux64.

  • Start the new version with: nohup ./spdb_2024A.1.1.1 > sysout.txt 2>&1 &

  • Here I tail the log (in log/spdb.0.0.log) to keep an eye on things, watching for the log message that the ODB is up. This should be quick since the database is empty. You may see stack traces as the seqexec hammers away trying to get an update.

  • Again, using the telent admin interface stop the trpc bundle to deny OT connections. It may no longer be bundle 70 so you have to hunt for it. (It is now #53 in this case.)

  • Stop the xmlrpc server: (edu.gemini.wdba.xmlrpc.server (now bundle #16 in my case).

  • Now it is time to import the XMLs that were exported earlier. Since I used the normal backup, I'll need to unpack the .zip file that it creates.

cd /mount/petrohue/odbhome/ugemini/spdb/spdb.archive/archive
unzip 20231212-0947.zip
  • With the telnet admin interface import the XMLs that were exported earlier: importXML /mount/petrohue/odbhome/ugemini/spdb/spdb.archive/archive/20231212-0947. This takes a long time. 30-45 min or so.

  • Here I need to remove the unzip'ed archive file (or your backup directory if you did it with exportXML) because the unzipped backup takes a lot of space on disk. rm -r /mount/petrohue/odbhome/ugemini/spdb/spdb.archive/archive/20231212-0947.

  • Start the trpc bundle and WDB server bundle (#53 and #16 respectively for me) and check the connection with the 2024A OT.

  • Edit the crontab again with crontab -e to turn on any cron jobs that were previously disabled.

  • If the daily report didn't run before you did the update, fire one off (see the command in the crontab).

  • Finally start the datamanager via the telnet admin: frameworklevel 100.