报错：
LSF提示userName <thesre>. getpwnam() failed:Bad user ID
原因：
由于LDAP在LSF进程之后启动，LSF无法获取到LDAP用户信息。参考：Jobs stuck pending on newly provisioned or diskless nodes using LDAP for user authentication. 
https://www.ibm.com/support/pages/jobs-stuck-pending-newly-provisioned-or-diskless-nodes-using-ldap-user-authentication
解决：
在出现这样问题的服务器，执行
badmin hrestart摘抄，以防止源丢失：
Jobs stuck pending on newly provisioned or diskless nodes using LDAP for user authentication.
Troubleshooting
Problem
When provisioning diskless nodes via Xcat or the first time provisioning a diskfull node which are LDAP clients for user authentication, jobs may fail to start.
Symptom
Jobs fail to start with errors in the sbatchd log like:
acctMapTo: No valid user name found for job <240>, userName <username>. getpwnam() failed: Bad user ID
Cause
getpwnam() fails when LDAP client is started later in the startup process (after LSF).
Environment
Xcat provisioning of x86_64 machines that set up LDAP client for user authentication
Diagnosing The Problem
The aforementioned error message in the sbatchd log indicates the problem.
Resolving The Problem
For machines already experiencing the problem, restarting the sbatchd will resolve the issue:
badmin hrestart
Any stuck jobs will have to be killed and resubmitted.
To prevent the problem, LDAP client configuration needs to occur before LSF configuration in the startup sequence.
For example, if you are using the setupcnldaplient script to configure LDAP client and starting up using a postscript, then the setupcnldaplient script must be listed before the LSF startup script in the postbootscripts list. . In the following example the image profile is named "rhels7.1-x86_64-stateless-compute "and the only other postbootscipt being used is called KIT_phpc_lsf_config, which configures LSF.
# chdef -t osimage -o rhels7.1-x86_64-stateless-compute \ postbootscripts=setupcnldaplient,KIT_phpc_lsf_config
There are other ways of ensuring the sequencing as well. In those cases the startup and configuration of the LDAP client must happen before the startup of LSF.
Document Information
More support for:
IBM Spectrum LSF
Software version:
10.1
Document number:
630603
Modified date:
17 June 2018
Need support?
- Submit feedback to IBM Support
-
1-800-IBM-7378 (USA)
- Directory of worldwide contacts
Contact IBM Privacy Terms of use Accessibility