Open danpovey opened 5 years ago
I have no bugfix, but maybe a hint how to get the system up and running again.
The SGE uses a database to store the runtime information. There are two possibilities: Files and BerkeleyDB.
Make a backup, before you peek around in the config!!!
The Files-DB can easily be edited and the nodes can be deleted manually with a text editor.
With BerkeleyDB you can use some of the tools which are installed together with SGE in "utilbin", or generic BerkeleyBD utils (usually it is hard to find the correct version).
Hope this helps to bring the SGE in a running state.
Thanks... I managed to fix my problem by deleting the jobs db after backing it up, restarting the master, re-adding the node whose deletion caused the problem, then stopping the master, copying back the old jobs db, and restarting the master. This would have confused clients, except I stopped their daemons first.
Not ideal, of course.
I believe the bug is in allowing a node with 'dr' jobs to be deleted. Shouldn't be hard to fix if someone knows the code.
On Thu, Nov 15, 2018 at 2:02 AM Marco Schmidt notifications@github.com wrote:
I have no bugfix, but maybe a hint how to get the system up and running again.
The SGE uses a database to store the runtime information. There are two possibilities: Files and BerkeleyDB.
Make a backup, before you peek around in the config!!!
The Files-DB can easily be edited and the nodes can be deleted manually with a text editor.
With BerkeleyDB you can use some of the tools which are installed together with SGE in "utilbin", or generic BerkeleyBD utils (usually it is hard to find the correct version).
Hope this helps to bring the SGE in a running state.
— You are receiving this because you authored the thread. Reply to this email directly, view it on GitHub https://github.com/son-of-gridengine/sge/issues/9#issuecomment-438938807, or mute the thread https://github.com/notifications/unsubscribe-auth/ADJVuySBX_P7tvjXBrziS00slcmcwVJHks5uvRF0gaJpZM4YYn4d .
See email chain pasted below. The basic issue, I believe, is that you can do qconf -de some_host when there are jobs in state 'dr' on that host. That crashes the gridengine master, and restarting it is not possible: message in /var/spool/gridengine/qmaster/messages is:
11/10/2018 16:23:27| main|deb8qmaster|C|!!!!!!!!!! got NULL element for EH_name !!!!!!!!!!
I'm not sure which part of the code deals with this; it should probably be fixed.