[Ovirt-devel] Re: qpid errors

[Perry Myers]

Ian,

Just checking status on some of the open issues that I had...

Perry Myers wrote:
> Perry Myers wrote:
>> You don't need to fix this tonight ;)
>>
>> But I figured I would just list the issues so that we tracked them 
>> somewhere.
>>
>> 1. fix race condition where it's possible that the broker gets the 
>> Node object before the db entry is created via ovirt-awake.  This can 
>> be fixed by having a hashtable of nodes and for each node a dirty 
>> flag.  If db-omatic goes to update the state in the db and can't it 
>> flags the node as dirty.  On the next heartbeat if the flag is dirty 
>> for a particular node it tries to update the db again.  The dirty flag 
>> isn't removed until the update succeeds.  This should only be a 
>> temporary hack until we get ovirt-awake using qmf.

This one you said is fixed correct?

>> 2. fix objects to be persistent so that restarts of the broker do not 
>> lose state

This was a non-issue, correct?

>> 3. get one of the ui guys to fix the MHz display so that the value 
>> there shows up correctly

sseago did this

>> 4. figure out why libvirt and hypervisor version are showing up as 
>> unknown

We need newer libvirt according to danpb (0.5.1)

>> 5. db-omatic is not logging properly

You fixed this

>> 6. rsyslog is not working on node startup...  you have top manually 
>> restart the rsyslog daemon after the node has booted to get a 
>> /var/log/messages file
>>
>> I'll take a look at #6, the rest are qpid related so they're yours...

I still need to figure this one out

> Oh adding to the list
> 
> 7. Get MRG to repackage their RPMs so that doxygen stuff is built during 
> RPM build and not as part of the source tarball and also get them to use 
> something other than 24 bit color for their png files.  And put their 
> docs into a docs rpm so that the devel rpm is not bloated.

Have you talked to the MRG guys about this?

Perry