Re: [rdiff-backup-users] Client dying "randomly" on 1.1.16

[Oliver Hookins]

On Tue Aug 26, 2008 at 15:11:26 -0400, Andrew Ferguson wrote:
> On Aug 21, 2008, at 6:46 PM, Oliver Hookins wrote:
>> Read from remote host exampleclient.backup: Connection timed out
>> Fatal Error: Lost connection to the remote system
>>
>> The two data files it accesses last are an 8KB and a 70MB postgres  
>> data
>> files. They don't seem out of the ordinary. From this trace I can't  
>> figure
>> out what is wrong. Maybe something here will trigger someone's memory.
>
> Oliver,
>
> If you're still trying to figure out why rdiff-backup is dying randomly, 
> here's another theory:
>
> Perhaps you are running rdiff-backup with a low priority? (eg, nice +10 
> rdiff-backup ...) In that case, my theory would be that some other  
> CPU-intensive (or HD-intensive) process is running at that time and  
> starves rdiff-backup while it is trying to compute the rolling checksums 
> for the rsync algorithm. In that case, rdiff-backup wouldn't send enough 
> data over the wire to prevent the connection from being dropped.

We run rdiff-backup with standard priority, but I think for whatever reason
it is getting held up on large files while lots of I/O is taking place. I
grabbed a tcpdump of the traffic and it looks like there is normal SSH
traffic until it hits whatever snag, then there are no packets for two
hours.

Then the connecting machine sends a TCP keep-alive, to which the machine
being backed up sends a TCP reset. So I think maybe my keep-alive settings
need some tweaking.

What sort of settings do people use in their ssh configs for BatchMode,
ServerAliveCountMax, and TCPKeepAlive?

-- 
Regards,
Oliver Hookins
Anchor Systems