On Sun, 31 May 2020 14:28:32 +0100, Adrian wrote:

FWIW, I see something fairly similar from time to time but its fairly
rare - I can go months without seeing it, and then I get one glitch.

My setup:

Hardwired network, all machines (two Lenovo laptops and an ancient dual
Athlon box, all running 64bit Fedora plus one series 2 RPi running
Raspbian Buster) are connected to a single D-link switch. My ADSL Modem/
router is also connected to the switch along with a M402dne Laserjet.

Backups are to a cycle of two of WD Essentials 1TB USB drives, stored
offline and connected to the dual Athlon box when I make a weekly backup
of all four systems immediately before a weekly software update. I use
rsync to do the backups, controlling and monitoring the process from a
SSH login on the Athlon box.

When a backup is in progress I see a progress report (i.e. files being
updated on the USB drives scroll up the console window that's logged in
to the Athlon. Sending copies of rsync are launched remotely from input
the the login session on the Athlon.

If the backup breaks its always because the sending copy of rsync that
has failed: I notice that the screen on the receiving end is no longer
scrolling up and, if I login to the sending system its always its rsync
instance that's crashed.

I learned long ago that this type of rsync backup arrangement is always
restartable without data loss (because the file that was being backed up
is the first to be backed up again when the run is restarted), so I
recover by killing the receiving copy of rsync (this forces a cleanup at
the receiving end) and then restarting that backup.  The only loss is of
time - the time it takes to notice that a backup has failed, clean up
and restart it.

Because recovery from such an infrequent failure is very fast, I will
admit to not investigating the causes in any detail.


--
Martin    | martin at
Gregorie  | gregorie dot org

--- SoupGate-Win32 v1.05