> We happen to have a failed SE at the moment. It's the 'alternate', > so no immediate impact at all. But the question has been raised, > what would happen if the primary failed also? I remember a case > years ago (1990s) when there was only one SE in a box, and it went > bad. Just meant that we could not IPL or make configuration changes > in the meantime, but it did not affect any running system. In those > days we had external timer boxes (9370?), so the sysplex was > unaffected. Now we have STP, which I believe is a component of the SE. > > If a modern box had no working SE for some period of time, would the > sysplex across multiple CECs be affected?
On the old bipolar machines, the SE was IPLed (which they referred to as a "warmstart") during each transition between Single Image and Physically Partitioned mode. MVS would hang when it tried to do things which synchronously interfaced with the SE while it was warmstarting. One thing which could cause this was an ACR. MVS used to spin waiting for the SE to deconfigure the ACRed CPU. We eventually changed the MVS code to do that asynchronously. SIGP Restart would also hang on the bipolar machines while the SE was warmstarting. The advent of LPAR reduced MVS's interactions with the SE, because LPAR virtualized the operating system's interface to the SE, and some of the interactions are now handled by LPAR with no SE involvement. It has been quite a few years since I have seen a dump of an MVS hang due to the SE being unavailable. Jim Mulder z/OS System Test IBM Corp. Poughkeepsie, NY ---------------------------------------------------------------------- For IBM-MAIN subscribe / signoff / archive access instructions, send email to [email protected] with the message: INFO IBM-MAIN
