The new version of restoreFromBackup: spins up X threads to speed up the operation, where X is the number of extents in the repository. Each of these threads are treated like a separate GemStone session, and will need to login and gain access to the shared page cache just like any other GemStone session.
If X exceeds the number of available sessions (limited by STN_MAX_SESSIONS) or the number of SPC slots (limited by SHR_PAGE_CACHE_NUM_PROCS), the operation will fail with the error:
-----------------------------------------------------
GemStone: Error Fatal
Error during login:
Error Category: 231169 [GemStone] Number: 4147 Arg Count: 0 Context : 20 exception : 20
Note that additional information is printed out in the gem log (or to stdout for a linked topaz session) that can indicate which parameter is causing problems.
If constrained by STN_MAX_SESSIONS you will see messages of the form:
PagesInOt-XX: Login error: Login failed: the maximum number of users is already logged in.
...
slave thread login error PagesInOt-XX: Login error: Login failed: the maximum number of users is already logged in.
If constrained by SHR_PAGE_CACHE_NUM_PROCS you will see messages of the form:
PagesInOt-XX: Login error: Unable to attach to shared cache
...
slave thread login error PagesInOt-XX: Login error: Unable to attach to shared cache
Note that most sites use the default value of -1 for SHR_PAGE_CACHE_NUM_PROCS, which calculates a reasonable value based on the setting of STN_MAX_SESSIONS, but which *might not* be large enough to cover the additional threads from restoreFromBackup:.
Depending on the error message seen in the gem log / linked topaz output, increase the setting of STN_MAX_SESSIONS or SHR_PAGE_CACHE_NUM_PROCS. If using the default setting of SHR_PAGE_CACHE_NUM_PROCS, you can just increase STN_MAX_SESSIONS rather than trying to calculate a reasonable value as described in the comments for SHR_PAGE_CACHE_NUM_PROCS in the configuration file.
Last updated: 12/5/14