FRS advanced questions

[Loekje]

Hi,

A client uses File Replication Services to synchronize two servers acros a DSL-VPN WAN.
This works fine, but we had to turn on FRS for about 100 directories one at the time:
The master has about 20 GB data, which was pre-copied with a portable harddisk to the "target". But this target has only a 36GB harddisk. Now we've found that *all* the files were moved to the "NTFRS_PreExisting.." directory but not directly moved back into the target directory. So you end up, temporarily, with twice the data (and twice the 20GB didn't fit on the 36 GB)

- Now is there a way to consolidate all the 100 directories back into the one directory of the original, or does that re-enter the stage of replicating the whole 20 GB?

- A related FRS question is about hanging files in the FRS-Staging directory, those files are a few days to weeks old. Obviously there is something wrong with them.
Can they be removed safely? Or do you break FRS with this?
Does a file that *is* yet to be replicated and is removed from the FRS-Staging replicated again?
If so than a scripted cleanup of old files would be no problem?

Awaiting great wisdom,
Loek Gijben

 

[SaraR]

I am also finding old files in the FRS-Staging directory, the contents of which seems to grow ever larger. We have already increased the maximum size several times....did you ever find an answer to whether it is safe to delete these files?

Thanks in anticipation,

Sara

 

[svsawkar]

>The master has about 20 GB data, which was pre-copied with a portable harddisk to the "target".

This is where things went wrong. FRS uses a combination of information to determine if a file is unique- one of which is a GUID. By per-copying the data, you changed the file GUID and all that new data came up as being 'new' files. Then FRS saw that you already had existing folders of the same name, so it renames collisions to the NTFRS convention.

Files in the Staging Directory are waiting to be sent to a replication partner- if they are stuck there, there may be another issue causing things to break.

Sara- what errors are you seeing in the FRS event logs?

/Siddharth

 

[SaraR]

The only event log error appearing is Error 13512 regarding "Enabled Disk write cache" which may cause problems in the event of drive failure, so I don't think that is the problem here.

To provide a few more details, there are 4 servers which are members of the replica set - servers 2,3 & 4 have 400, 22 & 5 files in FRS-Staging (obviously varies as files go through). Server 1 has 46,000 files which is taking up 5.5GB!! Something is evidently wrong. In the past we have had errors with the staging size, which has been increased to 8GB on all servers. This morning I have tried decreasing the staging area size on server 1 to 5GB to try and force DFS to clean iteself up. This has had no effect - I haven't even got messages to say it is full which surprises me.

Any thoughts would be appreciated!

Regards,

Sara

 

[svsawkar]

Stop and start the NTFRS server on both servers- you should be seeing more than just the disk caching message. Either a 13516 or a 13509. Lets see all of the messages after the service starts.

/Siddharth

 

[FrauW]

How about purging the staging area? I anticipate that the data will keep growing. Isn't there a way to make the staging area automatically delete files and create space for itself as needed. I have followed the Event Viewer suggestion to edit the Registry and increase the space however, I cannot increase the staging area space indefinitely? I also ran dsfgui.msc from Start->Run and took the replicas off- and online to 'jump start' them. However my data is still not replicating. Tips? About how long should I be giving the system to regulate itself?