RPC errors when provisioning schedule with many datasets

Question

When adding a dataset to a large schedule, we experience RPC errors as described below.
The target schedule is not changed.
The error appears when the dataset list gets to be around 10KB, e.g.
crbkpadm-sca[root]# wc CR-SCHED-BKP-INCR.lst
255 255 10652 CR-SCHED-BKP-INCR.lst
[ CR-SCHED-INCR.lst contains the dataset list for the schedule ]
They following illustrates the error:
/usr/bin/obtool chsched -A /HOSTS/sorrento-b/export/pool-1/cor-tools CR-SCHED-BKP-INCR
Error: couldn't update "CR-SCHED-BKP-INCR" - too many bytes supplied by RPC partner
where the schedule looks like
/usr/bin/obtool lssched -l CR-SCHED-BKP-INCR
CR-SCHED-BKP-INCR:
Type: backup
State: disabled
Dataset: /HOSTS/sorrento-a/export/pool-0/cdms_tools
/HOSTS/sorrento-b/export/pool-1/tahoe-lib-work2
/HOSTS/sorrento-a/export/pool-0/cr-db
/HOSTS/sorrento-a/export/pool-0/das-tools
/HOSTS/sorrento-a/export/pool-0/datools
/HOSTS/sorrento-a/export/pool-0/datools-rw
[ stuff deleted ]
/HOSTS/siena-b/export/pool-1/vt40-lib-ref
/HOSTS/siena-b/export/pool-1/vt40-lib-work1
Encryption: no
Comment: Incremental backup schedule CR-SCHED-BKP-INCR
UUID: b5732154-2be9-102e-9839-bb8b5939e4b0
Our setup is
o administration master - obtool version 10.3.0.3.0 (Solaris) -- Thu Aug 19 11:42:22 PDT 2010
o media server - obtool version 10.3.0.3.0 (linux) -- Thu Aug 19 06:33:20 PDT 2010
o SL8500 library - managed by ACSLS 8.0.2
Any ideas on how we can resolve this?
Thanks,
-pc

Answer 1

You need to file that with Oracle Support as it does sound like a bug.
However, splitting out your backups into individual datasets isn't the usual model. It's better to combine all similar items into a single dataset. I have some very large deployments with no more than 5 datasets per schedule. Typically I combine all similar clients into one dataset file, for example called LINUX and that contains many lines of Linux hosts.
You could also use nested datasets, so your schedule could call a single dataset called sorrento-a and that would "include dataset" to build up the rest of the entries you want. That way your schedule only has to have a few datasets, and you don't have to modify the schedule when you make changes, you just change your datasets.
So I would do this :
schedule calls datasets sorrento-a,sorrento-b,siena-a, siena-b
Dataset sorrento-a :
include dataset /HOSTS/sorrento-a/export/pool-0/cdms_tools
include dataset /HOSTS/sorrento-a/export/pool-0/cr-db
include dataset /HOSTS/sorrento-a/export/pool-0/datools
include dataset /HOSTS/sorrento-a/export/pool-0/datools-rw
Dataset sorrento-b :
include dataset /HOSTS/sorrento-b/export/pool-1/tahoe-lib-work2
Dataset sienna-b :
include dataset /HOSTS/siena-b/export/pool-1/vt40-lib-ref
include dataset /HOSTS/siena-b/export/pool-1/vt40-lib-work1
Thanks
Rich