I found a list of users I cannot add to an account: sacctmgr -i add user operator Accounts+=site_users Request didn't affect anything Nothing added. Make sure the accounts requested here have been added to the system/cluster. (sacctmgr add account...) To prove this account works: [root@cadence-rh76-lsf-1 tmp]# sacctmgr -i add user llawrenc Accounts+=site_users Adding User(s) llawrenc Settings Associations = C = slurmatnorth01 A = site_users U = llawrenc This is a valid account: [root@cadence-rh76-lsf-1 tmp]# id operator uid=11(operator) gid=0(root) groups=0(root) How do I debug why this account cannot be added?
(In reply to Larry Laugesen from comment #0) > I found a list of users I cannot add to an account: > > sacctmgr -i add user operator Accounts+=site_users > Request didn't affect anything > Nothing added. > Make sure the accounts requested here have been added to the system/cluster. > (sacctmgr add account...) > > To prove this account works: > > [root@cadence-rh76-lsf-1 tmp]# sacctmgr -i add user llawrenc > Accounts+=site_users > Adding User(s) > llawrenc > Settings > Associations = > C = slurmatnorth01 A = site_users U = llawrenc > > This is a valid account: > > [root@cadence-rh76-lsf-1 tmp]# id operator > uid=11(operator) gid=0(root) groups=0(root) > > How do I debug why this account cannot be added? Larry, This should be possible. I just confirmed that I could create this user and add it to an account my test system. Can you send me the output of the following commands on your system? uname -a sacctmgr show assoc tree Also, as a quick check, ensure the "operator" user exists on the controller and accounting machines and that the UIDs match. (if they aren't the machine mentioned above)
[root@cadence-rh86-spr-2 tmp]# id operator uid=11(operator) gid=0(root) groups=0(root) As for the association tree, this account has 16000+ users in it, is it possible I am running up against a size limit?
(In reply to Larry Laugesen from comment #2) > [root@cadence-rh86-spr-2 tmp]# id operator > uid=11(operator) gid=0(root) groups=0(root) I'm assuming you have centralized user management? LDAP? Are your controller and accounting machines the same? What backend are you using for slurmdbd? > As for the association tree, this account has 16000+ users in it, is it > possible I am running up against a size limit? I don't think that is the case, but we can test by adding a different user that doesn't already have an association. What O/S are you running? (The username and/or group name "operator" is special for some linux distributions)
In your original post, you mention a "list of users." What are the other users you are having trouble with?
The OS of the server in questions is Redhat 8.6 Below is some of the accounts: uucp sys pbctemgr sys tcinview
(In reply to Larry Laugesen from comment #5) > The OS of the server in questions is Redhat 8.6 > > Below is some of the accounts: > > uucp > sys > pbctemgr > sys > tcinview "operator," "uucp," and "sys" are special Linux accounts that shouldn't be used for slurm. Are slurmdbd and slurmctld both running on the same machine? Are you issuing the sacctmgr command on the same machine? If any of these are different machines, you need to make sure any accounts you want to add to slurm exists on each machine and has the same UUID. Is the RH 8.6 machine the same as the one mentioned above named "cadence-rh76-lsf-1"? Are you using LDAP or similar to sync accounts across machines?
