Hi John, We're not quite sure where the problem is coming from, but the discrepancies between ARSCHEMA numfields and the corresponding rowcounts in FIELD is an actual data discrepancy. The DB is not corrupted and appears quite readable, but ARS data (the field counts, the way views are displayed with missing fields) is definitely scrambled. To be clear, the data is scrambled as if someone was in there randomly running well-formed UPDATE statements and just inserting/modifying valid (but goofy) data in the meta-data tables. Actually, the last time, rows were missing from FIELD for HPD:HelpDesk and other forms (but the T,H and B tables all seemed ok).
Like I said, it's weird to the point of mind-boggling. To make matters worse the client's security restrictions place debilitating handicaps on trouble-shooting. They demand someone fix the engine, but refuse to allow anyone to open the hood (bonnet). So, I'm left with casting a wide net to see if anyone else has had or heard of similar probs... Thanks, JDHood On Sat, Oct 15, 2011 at 12:38 AM, John Peto <[email protected]>wrote: > Hi, > > Do you have affinity set between oracle client / db (think this is just tns > syntax), or are not using a load balanced VIP for the SID? TNS affinity > causes app servers to prefer a particular db node if it's available, can > make troubleshooting easier and would mean that all admin work is carried > out on one db node. SQL like the following show what is connected to what: > > set pagesize 1000 > column "From Machine" format a20 > select count(*) "Number of Connections", machine "From Machine", > b.instance_name > "To Instance" > from gv$session a, gv$instance b > where a.username='ARADMIN' and a.inst_id=b.inst_id > and machine in ('YOUR APP SERVER 1','YOUR APP SERVER 2') > group by machine, b.instance_name > order by "From Machine" > / > > Also not sure I understand where the problem is. Is the cache corrupted or > is the db corrupted? If the db is corrupted then the cache must get > corrupted 1st after a restore and then get commited back to db - by what > mechanism would that happen... Probably I'm just not understanding the > problem. > > It's such a crazy sounding problem that SAN cache or something might be the > problem, but that's crazy talk. > > Can't imagine any of this will be particulalry useful but i can't sleep :-) > > > _______________________________________________________________________________ > UNSUBSCRIBE or access ARSlist Archives at www.arslist.org > attend wwrug11 www.wwrug.com ARSList: "Where the Answers Are" > _______________________________________________________________________________ UNSUBSCRIBE or access ARSlist Archives at www.arslist.org attend wwrug11 www.wwrug.com ARSList: "Where the Answers Are"
