Loops after user disconnect and -519 error

[Lennart_T]

+ Retrospect version 7.6.123

+ Driver Update and Hot Fix, version 7.6.2.101

Clients either 7.6.107 or 7.7.114. This problem happens with both. I hoped that upgrading to 7.7.114 would fix the problem, but NO.

Disk backup sets, Proactive backup.

 

If a laptop user disconnects his/hers laptop during the "building snapshot" stage of the backup then sometimes the Retrospect server loops. The server uses 100% CPU and the other backups running slows considerably.

I can not stop this execution in any way (that I found). I click on the "Stop" button, confirms that I want to stop the execution and the "Stop" button is disabled. But the backup never stops. I have waited hours.

So I have to stop all the other executions (which takes some time due to CPU usage, but works out eventually.)

Then I have to "End task" in the Windows task manager.

I get the same error afterwards:

 

So:

Why is this happening?

How can I prevent it?

If I can't prevent it, how do I stop ONLY the offending execution?

 

What is the recommended client version for server 7.6.123?

 

Is this fixed in version 7.7?

[Richy_Boy]

I've not seen this exact issue in 7.7, although I do have backups hang on "Building snapshot" and have to crash retrospect in the same way to allow other clients to backup using the same script.

 

I really *really* hate cancel buttons which do nothing. It should try and cancel the script nicely until 'x seconds/minutes' and then kill the process if it fails to stop - rather than just sat hung blocking further backups.

 

Richard

[Mike Graham 95110]

I have a similar problem, with server version 7.6.11 and client v.7.6.106 ...yes, this is a very old version, from 2008 :-/  Right now I  have it running on Windows 7 SP1 with about 20 hosts.

 

I'll get the -519 Network Connection error, and the only way to get that particular backup to stop is to end the process in Task Manager.  CPU usage is nearly 100% until I end the process.  I am also curious if this is solved in 7.7.  It seems this issue would go away if Retrospect force-closed the problematic backup after X time, or offered an option to force-close after a user-defined time.   Our backups never take more than an hour, more typically they are 5 - 20 minutes.

 

 

I've been trying to chase down an issue with our network, where outbound access goes from 20 MBit (what we pay for) to 1 Mbit, meanwhile inbound is steady at or near 20 Mbit.   The link lights on our Cisco GB switches look like the lights on a hub with a ton of Ethernet collisions -- blinking so much they almost (but not quite) seem solid.  

This only seems to happen when a backup is in a "519" state, but even shutting off the server doesn't change the network issue, the network seems to clear of its own accord, and the server could display -519 for a day or two without the network issue showing up, or even appearing at all.   Unfortunately I don't have login access to the switches, firewall or router (managed by HQ IT) so I don't have any more information right now.  

 

Has anyone else observed this?  

am I on the right track, e.g. could Retrospect (or the network adapter) be sending spurious packets stemming from -519 Network Connection Error? 

Any other way to collect information, or possible fixes in case the Retrospect server is the offender?

 

[edited July 17 15:21 PDT]