‘GCR sleep’ event

Every database upgrade is a challenge for DBA to make it smoothly and without pain. Recently I did one, upgraded RAC database to + PSU5. After major post upgrade problems were fight I started to check performance, overall feeling was well but of course there were some bad SQL to be tuned and I noticed event I didn’t know yet and that was bothering me. This is how stats pack report look like:

Top 5 Timed Events                                            Avg %Total
~~~~~~~~~~~~~~~~~~                                           wait   Call
Event                                    Waits    Time (s)   (ms)   Time
--------------------------------- ------------- ----------- ------ ------
CPU time                                            21,587          44.9
db file sequential read              2,272,181       8,247      4   17.1
GCR sleep                                1,434       7,185   5010   14.9
db file scattered read                 629,679       2,865      5    6.0
enq: TX - row lock contention            3,291       1,786    543    3.7

And I couldn’t understand what is this sleep/wait event that occupies 15% of system resources and how I cannot see it from application performance or concurrent processing point of view . As neither MoS or google could help me SR was raised and this is what we get:

The GCR sleep event is an Idle Wait Event. This event should not be included by statspack when reporting on the Top Timed Events.
The following internal bug covers this issue:Bug 9695145 STATSPACK STANDBY: UPGRADE SCRIPT MISSING NEW IDLE EVENTS
This bug will be fixed in the next patchset release ( Please ignore the GCR sleep Event as it does not represent a performance issue.

So no worries, this is just a representative bug into Oracle Statspack.
How can I trust Statstpack now? :)


Output post processor monitoring

For starters quite a time since last post and it isn’t technical post, it is more my thoughts about topic.

Recently I got message from client to start monitor output post processor (OPP) concurrent manager and reason for that was that there were few accidents lately when OPP didn’t pick up request from or java garbage collector hung and failed OPP to operate. Either way there is need for DBA actions to bring services back. Reasons may be various for OPP to fail (I’ll not discuss the reasons) but as usually DBA is third level support there may be some time when You find out that system has problems.

Continue reading

Which responsibility can run this request?

Time to time when I’m doing some patching or configuration regarding documentation there is said: run request X with parameters Y or performing some configuration/upgrade tasks that automatically submits some requests but after review statuses there are some failed and need to be fixed and rerun. As apps DBA usually I have system administrator responsibility with SYSADMIN user account but eBS is built so that not always request is already assigned to system administrator responsibility and then I have edit request groups or add extra responsibility to SYSADMIN  role to be able to run the request. Of course  Oracle Forms can be used to perform search what is what but this can be very time consuming so I have written down some SQL that could save you some time.

Continue reading

Hardcoded path variables after RapidClone or Autoconfig

RapdClone and Autoconfig are power tools to manage eBS environments but however they’re not so mighty to manage all cloning/managing related things, especially when it comes down to customizations, development, etc. In this article I want to share my experience that I gathered during these years of cloning/maintaining/upgrading 11i/r12 environments for hardcoded path parameters into configuration. If you’re certain system apps dba for a long period of time then most probably you are very familiar with environment and this could be just a checklist but if you’re a newcomer or you have new client then during first clone I would suggest to check these very carefully .

Continue reading

NetAPP snapclone actual space allocated in storage

NetAPP provide huge advantage of administrate Oracle databases and advantages are even more if You’re using Snap Manager for Oracle. Only thing that I don’t like as DBA is that we usually don’t have access to filer and we have to ask some things to storage specialists.

When we’re managing volumes directly mounted from NetAPP they act pretty the same as NFS and we can use standard OS commands like du, df against volume to find out space usage, file sizes, etc. But still be very careful about results and always keep in mind if snapshots are taken to this volume then under hidden directory  .snapshot you’ll see files snapped and they can be easy managed as read only filesystem.

Continue reading

EBS DEV environment mass user disable/enable

Usually environment clones are used for any general purposes and sensitive data there are scrambled and as long as users are not complaining or trying to work with test instances instead of production there are no worries about access restriction. But time to time there is need for environment clone where data is not scrambled and then You have to think twice before give environment to end users. Best approach of course is restrict all and allow only  what is needed, so in this case we’re interested to disable log-in access to all users except ones who need that. Continue reading

R12 application server IP change side-effects

Few weeks ago there was issue when without DBA agreement system administrators changed eBS r12 application server IP address.  Regarding http://download.oracle.com/docs/cd/B32110_01/core.1013/b32196/host.htm#sthref405 this can be done and no any actions should be taken after, task is supported, everything should work after change. But it didn’t. Continue reading