Issue with PI 2.2 after upgrade

[patoberli]

Hello

I've upgraded today our PI 2.1 to 2.2. I followed exactly the manual and made an Application Backup no the old server, created a new server, installed it and restored the Application Backup.

Everything looked like it went well, but it seems I found a major issue now.

In the new Converged Theme -> Configuration -> Network Devices -> Unified AP, I see all my access points. But once I click on some device, I get the error message: 172.16.102.127 no longer exists. Please refresh the device work center.

Sometimes I can click on an access point, but it will show an other device in the overview, while the configuration seems to be from the correct one?!?

See screenshot:

I clicked on the 1200b-3103-1, but the title shows a different one. I didn't yet try to submit a change, because I simply do not trust it.

If I switch to the Classic (deprecated) theme, everything works as expected.

I've rebooted the server already twice, just to be sure.

Thanks
Patrick

[lbadman]

Count another happy site in on this bizarre bug, we're hitting it as well. Looking forward to spending weeks on the phone with TAC until we get some 12 year old DBA on the other side of the world that will tell us how this is a known bug that we're lucky to have.

PI is the Mayor of Suckville. It used to be on the Village Board, but now it runs the whole place.

[patoberli]

Quick update, my TAC is now 2 weeks open and they are currently looking into application backup that I've sent them.

Luckily the classic design works without any fault, so the priority for me isn't critical.

[patoberli]

Received an update. It seems I am hitting bug CSCup01845. TAC engineer will today try to make some SQL modifications on the server. I'll keep you updated.

[patoberli]

Positive news, the engineer could fix it.

I will post here the solution he had to do, but be aware that this might not work on your system!

This is only for 2.2.0.0.158 without any patches installed. You do this modification while PI is running. No restart required. Change is immediately active once you entered the commit.

It looks as this DB corruption happened in some older version while upgrading, maybe from 1.3 -> 2.0 or even earlier (original data is from WCS).

cpi1/admin# root
Enter root password :
Starting root bash shell ...
ade # /opt/CSCOlumos/bin/getDatabaseParams.sh
<Here_some_string_you_copy>
ade # su - oracle
[oracle@cpi1 ~]$ . ./oracleenv
[oracle@cpi1 ~]$ sqlplus / as sysdba

......

SQL> connect
Enter user-name: wcsdba
Enter password: <the_string_you_copied_earlier>
Connected.
SQL> update networkresource nr1 set name= (select wap.lradname from wirelessaccesspoint wap, networkresource nr2 where nr2.lradmneid = nr1.id and wap.wirelessaccesspoint_id=nr2.id) where nr1.owningentityid like '%UNIFIED_AP%' and exists (select * from wirelessaccesspoint wap, networkresource nr2 where nr2.lradmneid = nr1.id and wap.wirelessaccesspoint_id=nr2.id);

130 rows updated.

SQL> update networkelement ne set (mngmntaddrss_addresstype, mngmntaddrss_address)= (select wap.ipaddress_addresstype, wap.ipaddress_address from wirelessaccesspoint wap, networkresource nr where nr.lradmneid = ne.networkelement_id and wap.wirelessaccesspoint_id=nr.id) where ne.networkelement_id in (select managednetworkelement_id from managednetworkelement where owningentityid like '%UNIFIED_AP%') and exists (select * from wirelessaccesspoint wap, networkresource nr where nr.lradmneid = ne.networkelement_id and wap.wirelessaccesspoint_id=nr.id);

130 rows updated.

SQL> commit;

Commit complete.

SQL> quit

[Brian Sullivan]

You got way more information out of the TAC engineer than we got. All we could get was them identifying what our problem was. Reading the SQL statements above it looks like we hit something different. May be more than one bug. What we hit was this.

the client statisics is still in waiting state

In locked state.

Registered Locks
Lock NameMax LockersReference CountLocked By ThreadLocked ByLocked
AtLock TracebackThread WaitingWaiting InWaiting AtWait Traceback
PollSerializationLock11PollSerializationLock-1Task Serialization Lock(Client
Statistics)

Didn't tell us how they forced unlocked it.

However to your comments our database goes back to early WCS revisions and has been brought forward through every release since.

Thanks for the well documented steps to fix.

Brian