[Fedora-directory-users] Problem with Massive Replication Agreements

[David Boreham]

The message tells you that the server failed to create a thread
because the OS wouldn't let it. Typically this happens either because
there's some OS limit that needs to be increased, or because the
thread stack stride is very large and you're running a 32-bit server
(the fix for that is to reduce the stack size).

Fernando Muñoz wrote:
> Hi all,
>
> I have a distributed Directory Service on 25 servers with FDS 1.0.4. The
> replication configuration it's some complex:
>
> - One HUB server with 26 Databases(26 enabled Replicas)-> 25
> HUB-Replicas with 24 replication agreements and one Multimaster-Replica
> with 26 replication agreements. Total replication agreements in this
> server it's 626 replication agreements.
>
> - 25 Supplier (on our database) with Hub server and consummers (for
> others databases) of Hub server. 
>
> When i try to configure (automatically by script) this 626 replication
> agreements (on HUB server) i show this
> errorlog(/opt/fedora-ds/slapd-prueba/logs/errors) for each new
> replication agreement after 250 replication agreements aprox.:
>
> [22/Jul/2007:20:55:52 +0200] NSMMReplicationPlugin -
> agmt="cn=prueba1-prueba2" (prueba1:636): Unable to create protocol
> thread; NSPR error - -5974, Insufficient system resources.
>
> Exist some FDS limitation about massive replication agreements? or some
> FDS performance settings (maximun entries, caché,look-through limit,
> size limit, file descriptors,connection management...)?
>
> How can i solved this problem?
>
> thanks,
>
>
>
>  
>
> --
> Fedora-directory-users mailing list
> Fedora-directory-users at redhat.com
> https://www.redhat.com/mailman/listinfo/fedora-directory-users
>