Ciao, I wanted to prepare a docker container which serves as a disk-cached xrootd server, in order to deploy a T2-on demand (like) cluster.

I have used a simple configurations, attached in the end.
(By the way, I cannot find anywhere, do you know in which RPM it lives?)
The problem to me seems to come from the fact that in docker 'hostname' is not to be trusted; indeed whan I start xrootd I get

151110 09:08:27 166 XrdConfig: Unable to determine host name;  ai_family not supported ; execution terminated.

and the output of hostname -f is just

bash-4.1# hostname -f

Any way I can force xrootd to start in such situation?
thanks a lot



bash-4.1# cat xrootd-proxy.cfg 
# Tell everyone who the manager is
# Everyone will export /data red-only with the stage option. The stage
# option requests that if the file isn’t found in the cluster the
# redirector should send the client to a PFC server with enough space to
# cache the file.
all.export /store stage r/o
# First configure the redirector cmsd/xrootd using if-else-fi. However,
# the server cmsd configuration differs from that for the xrootd.
# So, we break those out in the if-else-fi clauses.
if redirector
all.role manager
# Server’s cmsd configuration – all PFC’s are virtual data servers
else if exec cmsd
all.role server
# The cmsd uses the standard oss plug-in but needs a specialized stat
# plugin to locate files in the cache. Also, the oss plug-in must be
# told where the cache is actually located via the space directive.
oss.statlib /usr/lib64/ public /data/xrootd-cache
# Server’s xrootd configuration – all PFC’s are virtual data servers
all.role server
# For xrootd, load the proxy plugin and the disk caching plugin.
ofs.osslib  /usr/lib64/ 
# Tell the proxy where the data is coming from (arbitrary).
# Tell the PFC’s where the disk cache resides (arbitrary).
pfc.cachedir /data/xrootd-cache 

Use REPLY-ALL to reply to list

To unsubscribe from the XROOTD-L list, click the following link: