Have a node that was added to HDP3.1 cluster via ambari and finding that the yarn user was assigned a uid that I need specifically for another user that we need existing across multiple cluster nodes (all having the same uid). After checking the other nodes in the cluster as well as a Hortonworks post, I see that the yarn user uids (as welll as for the various other service users) are randomly assigned on a per node basis. Eg.
[root@HW01 ~]# clush -ab id yarn
---------------
HW[01-04] (4)
---------------
uid=1019(yarn) gid=1005(hadoop) groups=1005(hadoop)
---------------
HW05
---------------
uid=1023(yarn) gid=1006(hadoop) groups=1006(hadoop)
Seeing this for many other users on node HW05 (note that host HW05 was added to the cluster via Ambari after the original cluster nodes were set up).
Is there any safe way (eg. I don't know any / all of the files and processes owned by the yarn user on any of the nodes) to change the uid of the (I assume ambari-assigned) uid of the yarn user (preferably via ambari)?
