Hi, Ran into an odd situation defining a storage box on spanned channels. There are 26 LPARs in this CPU (z15), 12 in LCSS 0 and 14 in LCSS 1, running z/OS 2.4. The storage box has 6 channel paths (01-06 FICON, spanned) directly connected (no switch) to this CPU, and there are 64 LCUs defined on the storage box. The limitation in the storage box is there can be no more than 1280 LCUs on any Host Adapter (read: CHPID).
In order to make this fit, in HCD we limit the paths from some LPARs in LCSS 0 (high IOPS LPARs get 6, lower IOPS LPARs get 3) in the access list, and this works out to 960 LCUs behind each CHPID, as expected. So far so good. The odd thing arises when we start limiting LPAR access in LCSS 1. Every CHPID sees 2 times as many LCUs on it than expected. For example, there are 3 LPARs that are limited to access CHPIDs 01, 02 and 03. I would expect that CHPIDs 01, 02 and 03 would each see 192 LCUs: 3 LPARs x 64 LCUs on each CHPID for 192 LCUs. What HCD reports is 384, not 192. All other LPARs in LCSS 1 also show twice what we expect, and this pushes us over the 1280 limit. When we reduce one of the 3-path LPARs on LCSS 1 from 3 paths to 1 path (changed access list from 04, 05, 06 to 06), the reduction should have been 64 LCUs on CHPIDs 04 and 05, but instead we saw a reduction of 128 LCUs, i.e. twice what we expected. To complicate things even more, channels 04 and 05 in LCSS 0 each got reduced by 64, but we didn't change anything in LCSS 0. Has anybody run into this sort of thing before? Gord Neill ---------------------------------------------------------------------- For IBM-MAIN subscribe / signoff / archive access instructions, send email to lists...@listserv.ua.edu with the message: INFO IBM-MAIN
