Did not die after sending SIGKILL

[GMRMacBackup]

Has anyone else experienced this issue? You ask the Retrospect Engine System Preference panel to 'Stop Retrospect Engine' and the process will not die? Even on good days it takes 3 to 5 minutes for the process to terminate, today it's been nearly 2 hours and still going strong. Attempting to stop the process again in the preference panel does nothing, I'd kill the process in Activity Viewer but the process is no where to be found.

 

4/29/10 6:16:38 PM com.apple.launchd[1] (com.retrospect.RetroEngine[3731]) Did not die after sending SIGKILL 7045 seconds ago...

4/29/10 6:16:43 PM com.apple.launchd[1] (com.retrospect.RetroEngine[3731]) Did not die after sending SIGKILL 7050 seconds ago...

4/29/10 6:16:48 PM com.apple.launchd[1] (com.retrospect.RetroEngine[3731]) Did not die after sending SIGKILL 7055 seconds ago...

4/29/10 6:16:53 PM com.apple.launchd[1] (com.retrospect.RetroEngine[3731]) Did not die after sending SIGKILL 7060 seconds ago...

4/29/10 6:16:58 PM com.apple.launchd[1] (com.retrospect.RetroEngine[3731]) Did not die after sending SIGKILL 7065 seconds ago...

 

I've heard the only way to kill this is to restart the server, but I hesitate to do this since it is a production server and I have users logged into it continuously.

 

System specifics:

• Intel Xserve (10.5.8 Server, 2 x 2Ghz, Apple Raid card, 3 x 750GB Raid 5, 10GB RAM)
• EMC Retrospect Single Server Unlimited 8.1.626
• Apple 4Ghz 4 Port FC Card (driver 2.02)
• Brocade 200e Fibre Channel switch (connected to Apple FC card)
• Promise 610f 9TB (Raid 5, 4GB FC controller, connected to Brocade)
• HP StorageWorks 4048 LTO4 dual drive w/4GB FC controller (connected to Apple FC card)

 

[breakaway]

The retroengine is controlled by launchd. To start and stop processes controlled in this way, you need to use the Terminal command launchctl. Here's what I enter in Terminal to stop and start the retroengine daemon:

 

To Stop:

launchctl unload /Library/LaunchDaemons/com.retrospect.launchd.retroengine.plist

 

To Start:

launchctl load /Library/LaunchDaemons/com.retrospect.launchd.retroengine.plist

 

I would try to locate the retroengine daemon in the Activity Monitor before doing this and noting the PID (process ID) of the daemon before and after to ensure that it restarting properly. If you don't have the retroengine daemon present before trying to stop it, you still need to issue the above mentioned command to "unload" it from launchd.

 

I also added these to my bash profile as aliases so I can just issue alias commands like "renginestop" and "renginestart" to make it easier to do this. I can explain that more if you're interested.

[GMRMacBackup]

Thanks for the terminal command, I'll keep them handy if the engine sputters on like an old clunker with a tankful of bad gasoline. Unfortunately it returned "Error unloading: com.retrospect.RetroEngine" and was unable to do anything. I think that was because the PID that is being reported by the Console log is not in the active PIDs in Activity Monitor. Looks like I'll need to chase everyone off of the server and restart to clear this error.

 

For the record, what caused this was selecting the 'Initialize elements' command from the pop-up menu of my tape drive that was offline. This is the second time activating this command has locked up the engine, guess I'll stay away from the command until a few more bugs are squashed.

[breakaway]

When I get "launchctl: Error unloading: com.retrospect.RetroEngine" while trying to run the stop command, I usually have to kill the RetroEngine process by using:

 

kill -9 PID (where the PID is the PID of the RetroEngine process as indicated in the Activity Monitor or top; and you'll need to run this as root)

 

Once that kills, you should be able to start the engine from either the PreferencePane or the start command mentioned above.

Edited April 30, 2010 by Guest