G
George Potemkin
Guest
Carl, can you save 'ipcs -s' output after each db startup? Footprint that the corruption is leaving in shared memory is the most painless way to find out the root case. The semaphore errors show what the processes found out in shared memory but we don't know what the same bytes stored after db startup. > Just had a call with Progress tech support, they suspect something is leaking file handles The size of the corrupted spot in shared memory, the byte alignment of the corrution's borders would confirm or deny the assumption. > server changed, from real redhat 5 to HyperV hosted virtual centos 7 I suspect this is the root.
Continue reading...
Continue reading...