Issues with Oracle on Solaris with Connections 2.5 UPDATE

As I posted on the Connections Blog earlier today …

At last it appears there may be a *real* fix for this .. IBM have changed the trigger code which should hopefully fix the mutating trigger issue we see when deleting files ..

I will be testing this today and if it works rolling it out to my live environment over the next week.

I will attempt to get some confirmation if this code will be fixed in all new fix packs and APARS – watch this space and hopefully some great news

At last ..

I am finally trying to sort out my website .. too lazy (and mainly lack of time) to be bothered to code up stuff or make it look pretty I have installed joomla, butchered one of the default skins and I am slowly loading it up with geeky goodness. you never know I may actually have a working website in a day or so .. but don’t hold me to that .. I am going back to work after a week off sick and have LOADS to catch up on .. but hey I am making a small effort πŸ™‚

Nerd Geek Dork

According to the diagram I am a Geek – which is reasuring as I was worried I was a Dork ..

This nerd/dork/geek/dweeb Venn diagram should save you a lot of time and frustration in the future.

this comment summed it up perfectly

The difference between Nerds and Geeks is that Nerds specialise and Geeks like diversity. If a Nerd has a favourite subject, they aim to make themselves the authority in it, whereas Geeks don’t take it that seriously – sure, its more serious than Average Person, but not Nerderious.

image from greatwhitesnark.com

JIRA 4.1 update SSL issues

after going nuts thinking that I had infact lost my marbles and had been following the instructions wrong I have discovered (thanks to Mr Andrew Frayling) that there is something missing from the Jira server.xml file.

A config changed happened between Tomcat 5.5 and 6 ( JIRA 4.0 uses 5.5, JIRA 4.1 uses 6 ) which means that you must have SSLEnabled=”true” in the secure connector port of the JIRA config – this is missing. Add it in and SSL suddenly starts working !!

http://jira.atlassian.com/browse/JRA-20963

It was drving me nuts all weekend!!

and now it works .. HURRAH πŸ˜‰

Issues with Oracle on Solaris with Connections 2.5 – UPDATE

After some testing with the SPARC version of this fix – which actually did work we were pleased to find out that Oracle had released a version for x86.

We applied this – this morning, and I am sorry to say it doesn’t work. If you try to delete a file from the DB directly or through the connections interface, the DB is still throwing the mutating trigger issue.

Plot thickens – time to go back to oracle πŸ™‚

Issues with Oracle on Solaris with Connections 2.5

There is an issue when running Connections with Oracle on Solaris
Symptoms of the problem are you can not delete certain files and / or the files widget from communities

The error in the logs is – table FILES.MEDIA is mutating

08/02/10 00:01:00:569 GMT] 0000005d Library E EJPVJ9166E: Unable to delete the library with id b855660b-d6bc-4b19-891f-2087aa3d9a0c. [UserImpl@26ce26ce id=64377ea3-e571-4323-922a-dc0723fead36 directoryId=2BE4B3FF-4AB4-48FF-9B83-73689537A16A]
java.sql.SQLException: ORA-04091: table FILES.MEDIA is mutating, trigger/function may not see it
ORA-06512: at “FILES.PKG_MED_DOWNLOAD_UPD”, line 45
ORA-06512: at “FILES.MED_DOWNLOAD_UPD_S”, line 2
ORA-04088: error during execution of trigger ‘FILES.MED_DOWNLOAD_UPD_S’

at oracle.jdbc.driver.DatabaseError.throwSqlException(DatabaseError.java:112)

We have since discovered (thanks to Kieran Reid in Connections Support for doing the leg work) that this is an issue with Oracle 10.2.0.4 on Solaris – the triggers have an issue which is fixed in 10.2.0.5 – which is a big no no as far as connections go. There is a fix that you can apply to 10.2.0.4 that will resolve the problem.

From support.oracle.com search the knowledge base for 4574851
You should get three results, select the third match
Click on the link for Patch.4574851
Select the 10.2.0.4 release for the Solaris platform
Download, install and test.

*NOTE* this fix is only available for SPARC not x86

So far this appears to have fixed the issue on the backup of the Prod database (I have put a stand-alone LC25 in front of it to test which involved all sorts of DB hacking to get it to work – not recommended unless you are desperate for a quick test). I am hoping to schedule moving our prod DB from x86 to SPARC applying the patch and then plugging my LC25 cluster into it.

Changing the title on the Connections 2.5 Homepage

Changing the title on the homepage is a bit of a pain .. The steps are as follows :

Make a back up copy of the COMPRESSED homepage.ear file from the deployed application config

ND – < was ROOT >profiles< profile> configcells< cell Name >applicationsHomepage.earHomepage.ear

StandAlone – < was ROOT >profiles< profile >configcells< cell Name >applicationsHomepage.earHomepage.ear

extract it to a temp folder ie D:tempextractedhomepage

find the dboard.common.jar and extract that to a temporary folder i.e D:tempextractedDashboard

drill down into the extracted file >

> com/ibm/lotus/connections/dashboard/nls/

and find the file jsp_resources.properties

change the jsp.homepage.title = < “your new title” >

change any instances of “IBM Lotus Connections Home Page” in this file to < “your new title” >

save and close the file

do the same for any additional languages that you are supporting

re-compress the dboard.common.jar and copy the newly edited compressed version into the extracted directory of the homepage ear file.

re-crompress the homepage ear file

stop all server instances that are running the homepage application replace the newly edited and compressed homepage.ear file in the deployed application config

you will also need to replace the newly edited dboard.common.jar in the installedApps folder on your primary / standalone server.

< was_root >profiles< profile name >installedApps< cell name >Homepage.ear

once the servers are restarted they will use the new title in the homepage app

Futher to the Portal 6.1.5 set up

Futher to the Portal 6.1.5 set up – these steps were required from an ORACLE DB point of view to get the DB configured :

Grant create session to community
Create likeminds user
Grant create table to community
Grant create view to jcr
Grant unlimited tablespace to community

and also these

grant select on sys.dba_pending_transactions to public;
grant select on sys.pending_trans$ to public;
grant select on sys.dba_2pc_pending to public;
grant execute on sys.dbms_system to public;

if you don’t do this then it just doesn’t work πŸ™‚

Portal won’t start ??

I have come across a horrible little *feature* that occurs sometimes with WebSphere Portal where the server fails to start and writes absoutely nothing to the log … Just a tad annoying when you are trying to work out why it didn’t start in the first place.

Sometimes this is due to a tranlog problem which is pretty straight forward to resolve:
backup / delete the directory –
< profile root >/tranlog/cellname/nodename/servername

normally this will do the trick and a restart works – if it doesn’t do the following
rename the log folders (or delete them)

< profile root >/logs

to fix the issue that I had seen I had to rename the ffdc, nodeagent and portal server log directories

restart the servers in question and as if by magic it starts πŸ™‚

I have experienced this at Portal 6.1.5 but I have had reports that this also appears to be a problem on 6.1.0.2 as well

Portal 6.1.5 do i love it or hate it ..??

Answers on a post card … first impressions of the installer was Wow that was neat, it installed the vanilla *everything* install fairly quickly and was up and running on it’s cloudscape db and basic security very quckly, looks funky, is fast and the new connections style theming is very sexy …

Now I am trying to do something wild and crazy …. point it at an oracle db … so far I am losing this battle .. although after a lot of swearing at windows 2008 for being TOO secure and the info centre for not making much sense I feel we may be winning the battle and be well on the way to winning the war, once the oracle piece is sorted it should be a straight forward secure against Active Directory federated LDAP cluster and go ..SHOULD – it was straight forward in Portal 6.1.1 but then so was securing it against an Oracle db …

Once I have it working I will doc the steps mainly for my reference but you never know some other poor little geek may also have to use Portal with the Oracle / AD solution and I would hate people to go through pain if they don’t have to πŸ™‚

watch this space kiddies you never know I may post something useful πŸ™‚