[Gluster-users] Transport Endpoint Not Connected When Writing a Lot of Files

[DUCARROZ Birgit]

Hi list,

Does anyone know what I can do to avoid "Transport Endpoint not 
connected" (and then to get a blocked server) when writing a lot of 
small files on a volume?

I'm running glusterfs 4.1.8 on 6 servers. With 3 servers I never have 
problems, but the other 3 servers are acting as HA storage for people 
who write sometimes a thousands of small files. This seems to provoke a 
crash of the gluster daemon.

I have 3 bricks whereas the 3rd brick acts as arbiter.


# Location of the bricks:
#-------$HOST1-------  -------$HOST3-------
# brick1            |  | brick3           | brick3 = arbiter
#                   |  |                  |
#-------$HOST2-------  --------------------
# brick2            |
#--------------------

Checked:
The underlying ext4 filesystem and the HD's seem to be without errors.
The ports in the firewall should not be the problem since it occurs also 
when the firewall is disabled.

Any help appreciated!
Kind regards,
Birgit