I have a problem regarding the EPO repository replication using the EPO 4.5 version. We are using 10 repositories. One of the repository is connected to the (W)LAN using a small bandwidth. It takes about 15 minutes tu copy 10 MB of data to the "problem" repository share. So the replication of all EPO data will take many hours. After one hour about 15% has been replicated. But after 5-6 hours the replication terminate with an error. The connection line is NOT broken (ping is working without errors).
Can it be, that there is a maximum replication duration parameter set?
Thanks and regards from Germany
I have contacted the McAfee support. They told me that the minimum bandwidth for replication is 512 KB/s. The used bandwidth is less than 512 KB/s.
Do you know a registry key to increase the timeout perid for replication? Please let me know this key. On which machine I have to configure this registry key? On the EPo server or on the destination repository server?
Thanks and regards
If you do not use global distributed repositories (that is, your repositories are defined in McAfee Agent policy rather than In ePO Menu - Software - Distributed Repositors) then you could consider using your own methods to make them up to date.
We use robocopy for keeping 20+ local repositories up to date (with 20+ individual Windows scheduled tasks) and its fast (can't say its going to be as fast in your case, but chances are its going to be faster than your current method).
What do you think?
AttilaMessage was edited by: apoling on 23/03/11 12:33:12 CET
about 50 clients - trend increasing...
We try to increase the bandwidth, but it will take some weeks. It would be fine, if we can temporarry set such a "non-limitation bandwidth registry key". But I don't know about this key.
With only 50 clients it would make more sense to simply allow them to update across the wan from the master repo (or another repo.) 50 incremental files each day are much, much smaller than replicating a full DAT package...
Yes that's one possibility. Bus as I wrote we will increase the numbers of clients and also try to increase the bandwidth step by step.
I’m interesting in that special "replication timeout registry key" – if such a key exists. Do you have information about such a key?
Bear in mind that you would need to have - broadly speaking - more than 400 machines on a client site before you need a distributed repository. (This is a very conservative estimate - some calculations put the number nearer 750.) Additionally, the lazy-caching feature in MA 4.6 (currently in beta) will hopefully solve the great majority of bandwidth-related replication problems: this may be available by the time the number of clients has increased.
If you really want to adjust the replication timeouts - which I recommend against - then, as Rob says, please open a case with Support, as these details are administered on a case-by-case basis.