[Freeipa-users] replication again :-(

[Ludwig Krispenz]

On 05/21/2015 02:20 PM, thierry bordaz wrote:
> On 05/21/2015 01:36 PM, Janelle wrote:
>>
>> And just like that - for no reason, they all reappeared:
>>
>> unable to decode  {replica 16} 55356472000300100000 55356472000300100000
>> unable to decode  {replica 23} 5545d61f000200170000 5552f718000300170000
>> unable to decode  {replica 24} 554d53d3000000180000 554d54a4000200180000
>>
>> :-(
>> ~J
>>
> Hello Janelle,
>
> Those 3 RIDs were already present in Node dc2-ipa1, correct ? They 
> reappeared on others nodes as well ?
> May be ds2-ipa1 established a replication session with its peers and 
> send those RIDs.
> Could you track in all the access logs, when the op 
> csn=5552f718000300170000 was applied.
>
> Note that the two hexa values of replica 23 changed 
> (5545d61f000200170000 5552f718000300170000 vs 5553e3a3000000170000 
> 55543240000300170000). Have you recreated a replica 23 ?.
>
> Do you have replication logging enabled ?
>
> thanks
> thierry
>
>
>
>
Hi Thierry, Mark,

I have an idea how this can happen, and now I have an environment where 
these show up.

The changelog contains max and purge ruv, and in my changelog I have:


dbid: 0000006f000000000000
         entry count: 304

dbid: 000000de000000000000
         purge ruv:
                 {replicageneration} 51dc3bac000000640000
                 {replica 100} 5555a759000000640000 5555a759000000640000
                 {replica 200} 5555b3c2000000c80000 5555b3c2000000c80000
                 {replica 300} 5555b3c20005012c0000 5555b3c20005012c0000

dbid: 0000014d000000000000
         max ruv:
                 {replicageneration} 51dc3bac000000640000
                 {replica 100} 5555a759000000640000 5555d773000000640000
                 {replica 200} 5555b3c2000000c80000 5555b3c2000000c80000
                 {replica 300} 5555b3c20005012c0000 5555b3c20005012c0000


after restarting I got:
  ldapsearch -LLL -o ldif-wrap=no -h localhost  -p 30522 -x -D 
"cn=directory manager" -w xxxxxx -b "cn=config" 
"objectclass=nsds5replica" nsds50ruv
dn: cn=replica,cn=dc\3Dexample\2Cdc\3Dcom,cn=mapping tree,cn=config
nsds50ruv: {replicageneration} 51dc3bac000000640000
nsds50ruv: {replica 100 ldap://localhost:30522} 5555a759000000640000 
5555d773000000640000
nsds50ruv: {replica 200 ldap://localhost:4945} 5555b3c2000000c80000 
5555b3c2000000c80000
nsds50ruv: {replica 300} 5555b3c20005012c0000 5555b3c20005012c0000

replica 300 is corrupted.

In this env I had played by cleaning ruv for rid 300, without disabling 
repl agreements from 300 (which I shoudl have done) and by adding 
changes later on replica 300 (which I shouldn't). Everything looked fine,
just after stopping to dump the changelog and restarting I was in the 
bad state

Need to try to repeat and verify








-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://listman.redhat.com/archives/freeipa-users/attachments/20150521/cd1a9fcf/attachment.htm>