Another one I categorize under 'I should write this down before I forget'.
- Oracle geek and entertainer extraordinaire - shared this with me last week. Apparently, he was having some memory issues after upgrading to 11gR2 under AIX. As we run 80% of our databases under AIX, I thought I should make a mental note about it (some may call it a blogentry).
For those with metalink access - lookup note id ID 1246995.1 titled 'Memory footprint for dedicated server processes more than doubled after 11g upgrade on AIX platform'.
The symptoms are :
- excessive memory utilization for dedicated server process after upgrade to 22.214.171.124 or 126.96.36.199 on AIX
- svmon on oracle OS process id shows size of USLA, User-Space Loader Assistant, heap of about 7M bytes
svmon -P <PID>
- 188.8.131.52.0 -> 7M bytes (=USLA x 4k page size)
- 184.108.40.206.0 -> 60KB
- 10.2.0.4.0 -> 420KB
Example on 220.127.116.11 (unpatched)
svmon -P 1368120 | grep USLA
10e811d 9fffffff clnt USLA text,/dev/hd2:8227 s 17 0 - -
10d89b9 80020014 work USLA heap sm 0 0 1895 1895 ---> x4K = 7Mb
Example on 18.104.22.168 (patched)
c291c2 80020014 work USLA heap sm 31 0 0 31 ---> x4K = 124K
This is apparently a bug
Bug 10211065.-P Base Bug 9796810
Abstract: MEMORY FOOTPRINT FOR DEDICATED SERVER PROCESSES MORE THAN DOUBLED AFTER 11G UGP
Bug 9796810.-P Base Bug 10190759
Abstract: 11G SERVER PROCESSES CONSUMING MUCH MORE MEMORY THAT 10G OR 9I
Abstract: PROCESSES CONSUMING ADDITIONAL MEMORY DUE TO 'USLA HEAP'
The good news is - there's a 'interim' patch - number 10190759.
It basically changes the make file so that it excludes certain options (-bexpfull and -brtllib) related to 11g's online patching feature.
It's not really an Oracle problem but it seems related to a bug in runtime loader on AIX.