Closed andreypz closed 6 years ago
Just do as normal user
eosfusebind
With normal user it does not work.
ls /eos/ ls: cannot open directory /eos/: Permission denied
It works for me
[mvala@lxplus002 ~]$ kdestroy
kdestroy: No credentials cache found while destroying cache
[mvala@lxplus002 ~]$ kinit
Password for mvala@CERN.CH:
[mvala@lxplus002 ~]$ eosfusebind
[mvala@lxplus002 ~]$ ls -al /eos/user/m/mvala
total 809K
drwx------. 1 mvala z2 35G Feb 7 21:57 .
drwxr-xr-x. 1 daemon root 66T Jun 22 08:57 ..
drwx------. 1 mvala z2 1.8G Jun 19 10:24 Documents
drwx------. 1 mvala z2 1.2G Jan 12 20:21 eBooks
-rw-r--r--. 1 mvala z2 405K Mar 18 2016 hsimple.root
drwx------. 1 mvala z2 27G Apr 18 09:05 Photos
drwx------. 1 mvala z2 53M Nov 13 2017 tmp
drwx------. 1 mvala z2 17M Mar 21 2017 tmva
drwx------. 1 mvala z2 4.4G Dec 15 2017 Videos
[mvala@lxplus002 ~]$ cat /eos/user/proc/whoami
Virtual Identity: uid=21495 (21495,99) gid=1395 (1395,99) [authz:krb5] host=lxplus002.cern.ch domain=cern.ch
[mvala@lxplus002 ~]$ id
uid=21495(mvala) gid=1395(z2) groups=1395(z2),1092772198 context=unconfined_u:unconfined_r:unconfined_t:s0-s0:c0.c1023
What does it mean "SLC6 AFS machine"?
On lxplus I don't have problems either (in fact, I don't need to run fusebind, it's mounted automatically).
What does it mean "SLC6 AFS machine"?
It is a non-lxplus machine at CERN, with SLC6 installed via http://linux.web.cern.ch/linux/install/
I see and did you configure eos fuse? can you show me files? /etc/sysconfig/eos.user
and /etc/auto.eos?
klist: No credentials cache found (ticket cache FILE:/tmp/krb5cc_0) could not find automatically a credential file
I think it's coming from eos. My ticket cache file in fact exists at /tmp/krb5cc_13052_D6hsrG
Hi Andrey,
I seems that eosfusebind
is not looking for the correct ticket cache. It's trying to read krb5cc_0
which is usually the ticket cache for root.
eosfusebind
(as do klist
and kinit
) tries to access the ticket cache pointed out by the KRB5CCNAME environment variable. It is set to root ticket cache usually when running su <user>
, this keeps the environment from root session.
To use a fresh environment, you need to run su - <user>
. Or just manually set the variable to the know ticket cache : export KRB5CCNAME=file:/tmp/krb5cc_13052_D6hsrG
So before running eosfusebind
, just try to run klist
and see what is the output of it. If it doesn't find a valid ticket, eosfusebind
will not work either.
Hi @mvala
My /etc/sysconfig/eos.user
is:
export EOS_FUSE_MGM_ALIAS=eoscms.cern.ch export EOS_FUSE_MOUNTDIR=/eos/cms export EOS_FUSE_USER_KRB5CC=1
The /etc/auto.eos
did not exist. I copied the file from lxplus machine, but it did not help.
@franck-eyraud : as I learned from @mvala the sudo is not necessary to run eosfusebind. So that is not the issue. When I run without sudo, there is no error. But simply the eos is not mounted.
So, the question here really is: how to mount eos on non-lxplus SLC6 machines. Do you have instructions for that?
I'm not a cern user, and we use EL7, but the procedure to mount a eos folder providing the file /etc/sysconfig/eos.user
is correct and you have root access should be service eosd start user
.
Otherwise, you should also be able to create a user mount with command eos fuse mount /home/user/mount-dir
, but I think this is not what you want to do.
Hi,
I'm on SLC6 AFS machine. When I do
sudo eosfusebind
, I get an error:I think it's coming from eos. My ticket cache file in fact exists at
/tmp/krb5cc_13052_D6hsrG