Restart LIM and mbatchd

[SASWayne]

I'm making a change to lsf.conf that requires to restart lim and mbatchd on grid control and candidate nodes. I've got to use "lsadmin reconfig" and "badmin mbdrestart" respectively for restarting lim and mbatchd.

Does this disrupt already existing jobs on SAS grid?
Also, what would be the impact on new job submissions during the restart happens?

Thanks!

[JuanS_OCS]

Hi @SASWayne ,

in theory it should not be much of an impact, specially for currently running jobs. The restart should take seconds or milliseconds, therefore the impact for new jobs it should not look worse than a small hiccup. But, of course, it all depends on several configuration and performance details.

[SASWayne]

Thank you for replying! @JuanS_OCS

I just want to make the changes I made to the lsf.conf effective, do you think using badmin reconfig instead of badmin mbdrestart would serve the purpose?

Thanks!

[JuanS_OCS · Accepted solution]

Excellent question @SASWayne ! It all depends if SDB needs to be restarted due to the changes on lsf.conf 🙂

See https://www.ibm.com/support/pages/difference-between-badmin-reconfig-and-badmin-mbdrestart

https://www.ibm.com/support/knowledgecenter/SSWRJV_10.1.0/lsf_admin/cluster_reconfig_lsadmin_badmin....

[SASWayne]

@JuanS_OCS Those links were really helpful! Thank you.

I'm going to reconfigure lsf.conf to make RES sessions to check SAS jobs status executing under it every X seconds. I believe that doesn't require to restart SBD after the reconfiguration. Let me know what you think please.

Thanks!

[JuanS_OCS]

I agree, I don't think this needs to restart SBD, give it a go 😉

[SASWayne]

@JuanS_OCS the new parameter didn't get effective on badmin reconfig, I've had to restart mbatchd to make it work.
Thanks for the LSF links!