rboltz Posted December 10, 2003 Report Share Posted December 10, 2003 I am currently using a PowerMac G4 running OSX 10.2.4 with Retrospect Server 5.0.238. I back up 100-150 Clients who have either a laptop or desktop running Mac OS9 or OSX all with Restrospect client 5.0.x. Since I have so many clients, I run most of my scripts in the evening. On occasion I will come in the next morning to find that the script started the day before is hung up on a client (not one particular, it varies) giving a "Net Retry" message. Everything I've read is that Net Retry is normally a communication error. There are two things I'm trying to figure out... 1. I've tested the communication lines between the server and the client(s) through pinging. Could the communication error occur if the energy saver puts the client's hard disk to sleep after a certain period of time? If so, what is the optimal setting for this? 2. Is there a way to have the retrospect server skip over the client in the event a net retry occurs? That way the remaining clients can still get backed up. The server will skip over a client if the client is not visible, but in these instances, the client is half-way backed up when the communication error occurs. Thanks for any input. Bob Link to comment Share on other sites More sharing options...
RaoulSevier Posted December 10, 2003 Report Share Posted December 10, 2003 What a coincidence. I was just about to post a note on this very topic. I find Retrospect does not time-out but stays in perpetual 'net retry'. This occurs when backup of a laptop begins but then is interrupted by being put to sleep (cover is closed). That client action causes Retrospect to 'net retry' and consume 12x cpu on the server (60% instead of the usual 5%). This is how I can tell this has happened: Server Admin's CPU use graph levels out at 70% perpetually instead of the usual 10%. And I generally discover this upon return from a trip only to find *all* backup activity has not taken place because it is looking for a laptop that left the network. Is there a way to time-out the 'net retry' activity after a few minutes? Thanks Link to comment Share on other sites More sharing options...
RaoulSevier Posted December 10, 2003 Report Share Posted December 10, 2003 Ah. Just found a reference to the 'hidden option' allowing termination of a specific backup based on transfer performance. This should do it for me. http://www.dantz.com/index.php3?SCREEN=kbase&ACTION=KBASE&id=28017 Link to comment Share on other sites More sharing options...
RaoulSevier Posted December 11, 2003 Report Share Posted December 11, 2003 Oops. My version (5.1 Server Mac) does not have a 'Client' entry in the list on the left side of the preferences panel. Now what? Thanks, Link to comment Share on other sites More sharing options...
RaoulSevier Posted December 11, 2003 Report Share Posted December 11, 2003 Option-preferences. All set now. Thanks. Link to comment Share on other sites More sharing options...
CallMeDave Posted December 11, 2003 Report Share Posted December 11, 2003 Quote: All set now. Thanks. Be aware that this a time-out with this setting will end the entire script, even if there are other Sources in line. For best results, the client speed threshold should be used with a Backup Server script; in this mode, each new Source behaves as a new script. If one client fails the test, Retrspect will proceed to the next source. Dave Link to comment Share on other sites More sharing options...
rboltz Posted December 13, 2003 Author Report Share Posted December 13, 2003 Any suggestions on an ideal Speed Threshold? I have mine set for 50k bytes/second and am still having Net Retries on clients. According to the description of this feature, this setting is to "Ensure the client connection exceeds this throughput before backing it up". For me, the net retry message is coming up in the middle of backing up a client. Link to comment Share on other sites More sharing options...
CallMeDave Posted December 14, 2003 Report Share Posted December 14, 2003 Quote: According to the description of this feature, this setting is to "Ensure the client connection exceeds this throughput before backing it up". That is indeed the description of a script's Client Execution _option_ (not available in Immediate Execution). This option has been in Retrospect for ages. The new preference introduced in 5.1 is described in the KB article referenced above. It is for all exectuions, so it's in Retrospects global prefs (although hidden in the secret preference window; Option+Click to view). This preference checks on Client connection speeds throughout the connection (for both Backups and Restores). Dave Link to comment Share on other sites More sharing options...
RaoulSevier Posted December 20, 2003 Report Share Posted December 20, 2003 Dave, thanks for the advice about using this feature with a Backup Server script. That is how I'm using it, having set the threshold to 1MB/Min. Just checked the server to find that CPU has been steady at 70% overall for the last 22 hours instead of the usual 10-15%. Looking at the Activity Monitor shows that Retrospect consumes 80% CPU for about 4 seconds then drops off the list for 2 seconds. This corresponds to the appearance of the Net Retry panel for 4 seconds and its disapearance for 2 seconds. The panel is looking for a laptop that went to 'Ready' status but then was taken off the network. That was 22 hours ago and ever since 60% of the server's CPU has been consumed in this Net Retry activity. Any suggestions for avoiding this situation? Link to comment Share on other sites More sharing options...
Recommended Posts
Archived
This topic is now archived and is closed to further replies.