M
Mycroft Holmes
Guest
Hi,
sorry for cross-post, but it seemed appropriate.
I'm trying to describe a problem for which I'm not totally familiar, so I
admit my message could be slightly imprecise in terminology.
we have a large c++ projects that compiles a windows service (and its
dll's), that runs on some big 8-way opteron servers.
On each server there's a single executable for the service, but it's been
registered twice i.e. there are entries in the registry SERVICE_A and
SERVICE_B that refer to the same exe, with different configurations.
The servers reboot once a day and sometimes on reboot, services.exe crashes,
leaving the whole machine in a zombied state.
1) this does never occur if SERVICE_B is removed
2) this seems to occur ONLY on reboot: we stress-tested the servers using
multiple simultaneous "net stop/net start", and all is fine.
we assumed that, even by mistake it should be almost impossible to crash
services.exe. does anybody have hints?
thanks in advance,
MH
sorry for cross-post, but it seemed appropriate.
I'm trying to describe a problem for which I'm not totally familiar, so I
admit my message could be slightly imprecise in terminology.
we have a large c++ projects that compiles a windows service (and its
dll's), that runs on some big 8-way opteron servers.
On each server there's a single executable for the service, but it's been
registered twice i.e. there are entries in the registry SERVICE_A and
SERVICE_B that refer to the same exe, with different configurations.
The servers reboot once a day and sometimes on reboot, services.exe crashes,
leaving the whole machine in a zombied state.
1) this does never occur if SERVICE_B is removed
2) this seems to occur ONLY on reboot: we stress-tested the servers using
multiple simultaneous "net stop/net start", and all is fine.
we assumed that, even by mistake it should be almost impossible to crash
services.exe. does anybody have hints?
thanks in advance,
MH